Docstoc

Document template for NOBEL

Document Sample
Document template for NOBEL Powered By Docstoc
					                      IST IP NOBEL "Next generation
                                                                  Title of the document
                      Optical network for Broadband
                           European Leadership"              e1686d81-2994-497c-93f6-
                                                      b2e913b07f59.docD06_DRAFT_V1
                                                                  6JW2 Andreas zweiter
                                                           Teil.docD06_DRAFT_V15ejw




                                                                                      Formatted: Bullets and Numbering
1Deliverable 6
1Work Package 1
      1“Preliminary definition of drivers and
    requirements for core and metro networks
supporting end-to-end broadband services for all”

Status and Version:       Draft
Date of issue:            01.03.2004
Distribution:             Project Internal
Author(s):                Name                          Partner
                         Anders Berntson               ACREO
                         Claude Artigue                Alcatel CIT
                         Bela Berde                    Alcatel CIT
                         Emmanuel Dotaro               Alcatel CIT
                         Miquel Huguet                 CESCA
                         Hisao Nakajima                France Telecom
                         Arie de Heer                  Lucent
                         Giovanni Fiaschi              Marconi
                         Vivek Kulkarni                Siemens
                         Antonio J. Elizondo           Telefonica I+D
                         Gabriel Moreno                Telefónica I+D
                         Jesús F. Lobo                 Telefónica I+D
                         Hans Carlden                  Telia-Sonera
                         Björn Hermansson              Telia-Sonera
                         Andrea Di Giglio              TIlab
                         Antonio Manzalini             TIlab
                         Nicolai Leymann               T-Systems
                         Rüdiger Kunze                 T-Systems
                         F.-Joachim Westphal           T-Systems
                         Josep Solé-Pareta             UPC




                               Page 1 of 168
              IST IP NOBEL "Next generation
                                                          Title of the document
              Optical network for Broadband
                   European Leadership"              e1686d81-2994-497c-93f6-
                                              b2e913b07f59.docD06_DRAFT_V1
                                                          6JW2 Andreas zweiter
                                                   Teil.docD06_DRAFT_V15ejw




Checked by:      WP1 reference person          All partners




                      Page 2 of 168
                                 IST IP NOBEL "Next generation
                                                                              Title of the document
                                 Optical network for Broadband
                                      European Leadership"                e1686d81-2994-497c-93f6-
                                                                   b2e913b07f59.docD06_DRAFT_V1
                                                                               6JW2 Andreas zweiter
                                                                        Teil.docD06_DRAFT_V15ejw




Table of Contents
1   Introduction                                                                               87
    1.1      Purpose and Scope                                                                 87
    1.2    Reference Material                                                                  87
       1.2.1 Reference Documents                                                               87
       1.2.2 Acronyms                                                                          98
    1.3      Document History                                                                1211
    1.4      Document overview                                                               1311
2   Extended Management Summary                                                              1513
    2.1      Definition of the Context                                                       1513
    2.2    Emerging Applications                                                             1715
       2.2.1 Definitions and goal of the research                                            1715
       2.2.2 Parameters characterizing emerging applications                                 1815
       2.2.3 Conclusions                                                                     2017
    2.3    Traffic Prediction and Characterizations                                          2118
       2.3.1 Traffic typology and relations concerning core and metro networks               2118
       2.3.2 Criteria for traffic classification                                             2118
       2.3.3 Traffic classes                                                                 2219
       2.3.4 Traffic characteristics per application and per user                            2219
       2.3.5 Impact of traffic on the network architecture                                   2320
        2.3.5.1    Network switching paradigms                                        2320
          2.3.5.2   Topology of the network                                           2320
          2.3.5.3   Connection dynamics                                               2320

    2.4    Network Modes and Services                                                        2420
       2.4.1 Network Modes                                                                   2420
       2.4.2 Network Services                                                                2521
        2.4.2.1  Network service table                                                2522
      2.4.3 Mapping of applications to services                                              2622
      2.4.4 Quality of Service characteristics                                               2723
      2.4.5 Concepts of VPN service and VPN trasport layers                                  2723
    2.5      High-Level Network Requirements                                                 2824
    2.6      Functional Modeling of the Transport and Service Platforms                      2925
          2.6.1.1   Modeling of Connection oriented layer networks                    3025
          2.6.1.2   Modeling of Connectionless layer networks                         3026

    2.7      Methodology for defining migration paths                                        3026
3   Major drivers for the evolution of existing core and metro networks                      3227



                                         Page 3 of 168
                                  IST IP NOBEL "Next generation
                                                                             Title of the document
                                  Optical network for Broadband
                                       European Leadership"              e1686d81-2994-497c-93f6-
                                                                  b2e913b07f59.docD06_DRAFT_V1
                                                                              6JW2 Andreas zweiter
                                                                       Teil.docD06_DRAFT_V15ejw




4   Emerging applications                                                                   3630
    4.1    Storage applications                                                             3630
       4.1.1 General Issues                                                                 3630
       4.1.2 Parameters                                                                     3832
        4.1.2.1   Resilience                                                         3832
       4.1.2.2    Provisioning                                                       3933
       4.1.2.3    Dynamics                                                           3933
       4.1.2.4    Bit-Rate                                                           4034
       4.1.2.5    Quality of Service (QoS)                                           4134
       4.1.2.6    Connectivity – Transparency – Interface                            4135
       4.1.2.7    Security                                                           4236
      4.1.3 Conclusions on Storage Applications                                             4336
      4.1.4 Important proprietary storage applications: IBM GDPS and EMC SRDF               4337
    4.2    Multimedia applications                                                          4437
       4.2.1 Background                                                                     4437
      4.2.2 Video streaming (video-on-demand, VoD, video broadcast, and IP-TV)              4740
      4.2.3 Voice over IP & video chat                                                      5142
      4.2.4 Video Conference                                                                5344
      4.2.5 Data applications (in particular video download)                                5445
      4.2.6 On-line gaming                                                                  5545
      4.2.7 Tele-medicine applications                                                      5646
       4.2.7.1  Tele- (or remote-) diagnostics                                       5747
       4.2.7.2    Remote (tele-) surgery                                             5747
       4.2.7.3    Medical data storage                                               5848
       4.2.7.4    Paramedic, emergency communications                                5848

    4.3    Grids                                                                            5848
       4.3.1 Definition                                                                     5848
        4.3.1.1  Compute Grids                                                       6050
       4.3.1.2    Data Grids                                                         6050
       4.3.1.3    Utility Grids                                                      6050
      4.3.2 Service Requirements                                                            6150
       4.3.2.1  Grid Scenarios                                                       6150
      4.3.3 Parameters                                                                      6252
       4.3.3.1  Resilience                                                           6252
       4.3.3.2    Provisioning                                                       6353
       4.3.3.3    Dynamics                                                           6353
       4.3.3.4    Bit-Rate                                                           6453
       4.3.3.5    QoS                                                                6454



                                          Page 4 of 168
                                   IST IP NOBEL "Next generation
                                                                              Title of the document
                                   Optical network for Broadband
                                        European Leadership"              e1686d81-2994-497c-93f6-
                                                                   b2e913b07f59.docD06_DRAFT_V1
                                                                               6JW2 Andreas zweiter
                                                                        Teil.docD06_DRAFT_V15ejw




          4.3.3.6    Connectivity – Transparency – Interface                          6655
          4.3.3.7    Security                                                         6655

    4.4      Summary of emerging applications                                                6756
5   Network Services                                                                         7159
    5.1      Introduction                                                                    7159
    5.2    Network service groups                                                            7259
       5.2.1 L3 VPN                                                                          7360
       5.2.2 L2 VPN                                                                          7461
       5.2.3 L1 VPN                                                                          7764
    5.3    Network service table                                                             8369
       5.3.1 Performance parameters                                                          8369
        5.3.1.1  Protocol transparency                                                8369
          5.3.1.2    Standards - VPN identifier                                       8369
          5.3.1.3    Provisioning                                                     8369
          5.3.1.4    Blocking probability                                             8470
          5.3.1.5    Configuration dynamics                                           8470
          5.3.1.6    Connectivity                                                     8470
          5.3.1.7    Service Availability                                             8470
          5.3.1.8    Bit-rate                                                         8470
          5.3.1.9    QoS                                                              8571
          5.3.1.10      Customer insight                                              8571
          5.3.1.11      Billing                                                       8571
          5.3.1.12      Security                                                      8571

    5.4    Mapping network services groups to applications                                   8672
       5.4.1 Storage applications                                                            8672
       5.4.2 Grid                                                                            8773
       5.4.3 Multimedia                                                                      8773
6   Traffic typology and relations concerning core and metro networks                        8874
    6.1      Criteria for traffic classification                                             8874
    6.2    Traffic classes                                                                   9379
       6.2.1 Inelastic traffic                                                               9480
        6.2.1.1    Real time applications                                             9480
          6.2.1.2    Streaming applications                                           9681
       6.2.2 Elastic traffic                                                                 9682
    6.3    Traffic characteristics per application and per user                              9883
       6.3.1 Metro Networks                                                                  9883




                                              Page 5 of 168
                               IST IP NOBEL "Next generation
                                                                              Title of the document
                               Optical network for Broadband
                                    European Leadership"                  e1686d81-2994-497c-93f6-
                                                                   b2e913b07f59.docD06_DRAFT_V1
                                                                               6JW2 Andreas zweiter
                                                                        Teil.docD06_DRAFT_V15ejw




       6.3.2 Core Networks                                                                 10187
       6.3.3 User profile                                                                  10287
    6.4    Impact of traffic on the network architecture                                   10389
       6.4.1 Network switching paradigms                                                   10389
       6.4.2 Topology of the network                                                       10490
       6.4.3 Connection dynamics                                                           10591
7   Emerging requirements for metro and core networks                                      10793
    7.1    Introduction                                                                    10793
    7.2    Architecture                                                                    10995
       7.2.1 Functionally separate architecture                                            10995
       7.2.2 Applications independent architecture                                         11095
       7.2.3 IP infrastructure and migration issues                                        11095
    7.3    Network requirements                                                            11196
       7.3.1 Data plane requirements                                                       11196
       7.3.2 Routing and Resilience                                                        11196
       7.3.3 QoS Requirements                                                              11297
       7.3.4 Traffic Management Requirements                                               11297
       7.3.5 Security and Authentication Requirements                                      11499
       7.3.6 Addressing requirements                                                      115100
       7.3.7 Billing and charging Requirements                                            116100
       7.3.8 Criteria for network evolution requirements                                  119104
       7.3.9 Control plane requirements                                                   120105
       7.3.10 Interoperability and Interworking Requirements                              123108
       7.3.11 Management plane requirements                                               124109
    7.4    General                                                                        126111
       7.4.1 Open Networks                                                                126111
       7.4.2 Price reduction on transport                                                 127111
    7.5    Service Requirements                                                           127112
       7.5.1 General                                                                      127112
       7.5.2 QoS                                                                          129114
8   Conclusions                                                                           131115
9   Appendix A: Definition and classification of transport layer networks                 132116
    9.1    Connectionless and connection oriented layer networks                          132116
10 Appendix B: Functional modelling of transport layer networks                           133117
11 Appendix C: Standardisation Bodies and Fora working on architectural
requirements and functional models for core and metro optical networking                  134118
    11.1 ITU-T Project Next Generation Network                                            134118
       11.1.1 Basic characteristics of NGN                                                134118
       11.1.2 NGN capabilities                                                            135119
       11.1.3 General architectural principles for the NGN                                136120
       11.1.4 Functional architecture methodology model                                   136120
       11.1.5 End-to-end Quality of Service (QoS)                                         136120
       11.1.6 Service platforms (APIs)                                                    137121



                                        Page 6 of 168
                           IST IP NOBEL "Next generation
                                                                             Title of the document
                           Optical network for Broadband
                                European Leadership"                     e1686d81-2994-497c-93f6-
                                                                  b2e913b07f59.docD06_DRAFT_V1
                                                                              6JW2 Andreas zweiter
                                                                       Teil.docD06_DRAFT_V15ejw




     11.1.7 Network management                                                          137121
     11.1.8 Security                                                                    137121
     11.1.9 Generalized mobility                                                        138122
     11.1.10    Network control architecture(s) and protocols                           139123
     11.1.11    Service capabilities and Service Architecture                           140123
     11.1.12    Interoperability of Services and Network in NGN                         140124
     11.1.13    Numbering, naming and addressing                                        140124
      11.1.13.1    General                                                         140124
     11.1.13.2     Achieving a Seamless User Experience                            141125
     11.1.13.3     Fundamental Principles for Name Resolution                      141125
     11.1.13.4     Routing Protocols                                               142126
     11.1.13.5     Related Standards                                               142126

  11.2   ITU-T SG13 (2001-2004)                                                          144127
  11.3   ITU-T SG15 (2001-2004)                                                          144128
  11.4   OIF                                                                             146129
  11.5   IETF                                                                            146129
12 Appendix D: Transparent Optical Network and Wavelength Services                       148130
  12.1   Purpose, Scope and document overview                                            148130
  12.2   Introduction                                                                    148130
  12.3   Transparent Optical Network                                                     149131
  12.4   Static & Dynamic Transparent Optical Networks                                   149131
  12.5   Switched Transparent Optical Network                                            152133
  12.6 Wavelength Service                                                                155136
     12.6.1 Wavelength Service over a Transparent Optical Network                        155136
     12.6.2 Wavelength Service over a Switched Transparent Optical Network               156137
  12.7   Switched Wavelength Service over a Switched Transparent Optical Network         157138
13 Appendix E: Applications questionnaire                                                160141
  13.1   Introduction                                                                    160141
  13.2   Results                                                                         162143




                                       Page 7 of 168
                             IST IP NOBEL "Next generation
                                                                                Title of the document
                             Optical network for Broadband
                                  European Leadership"                    e1686d81-2994-497c-93f6-
                                                                   b2e913b07f59.docD06_DRAFT_V1
                                                                               6JW2 Andreas zweiter
                                                                        Teil.docD06_DRAFT_V15ejw




 1         Introduction

 1.1       Purpose and Scope
The working group responsible for this deliverable is aiming at identifying network solutions
and evolutionary guidelines for core and metro optical/networks to support end-to-end
broadband services for all. The deliverable “Preliminary definition of drivers and
requirements for core and metro networks supporting end-to-end broadband services for
all” will be the documentation of the first steps in this direction.
This deliverable will preliminarily identify and investigate on the major drivers moving the
evolution of current transport networks. Particular attention will be paid to emerging
applications that are fuelling the growing of data-traffic such as Storage Area Networking
(SAN) for outsourced access/management of server content, disaster recovery services,
common access to remote resources, network-wide computation and data services, display
of multi-media information through virtual reality, etc.
Starting from those drivers the emerging requirements for core and metro networks
supporting end-to-end broadband services for all will be identified as well as other open
issues to produce a basis to find network solutions overcoming the current bottlenecks.


 1.2       Reference Material

          1.2.1 Reference Documents
[nobel1]       NOBEL: Deliverable 2 of Work Package 4
               “Definition of Network Management and Control requirements of network scenarios and
               solutions supporting Broadband Services for All”
[Berman]       Berman F, Fox G.C., Hey A. J.G., “Grid Computing – making the global
               infrastructure a reality”, Wiley series in Communications networking & distributed
               systems, 2003
[Abbas]        Ahmar Abbas, “Grid Computing: A practical guide to technology and applications”,
               Charles River Media, inc. 2004
[Gara]         http://www-fp.mcs.anl.gov/qos/gara.htm
[ICICS]        http://www.singaren.net.sg/activity/ICICS03.pdf
[Skanova]      Product offers by Skanova, a branch of Teliasonera, www.skanova.se (in
               Swedish).
[EBU I34]      EBU Technical Information I34-2002
[EBU I35]      EBU Technical Information I35-2003
[ETSI1]        Draft ETSI, Digital Video Broadcasting (DVB) DVB-IP Phase 1 Handbook
[ETSI2]        ETSI TS 101 329-2 v2.1.3




                                       Page 8 of 168
                                  IST IP NOBEL "Next generation
                                                                                                 Title of the document
                                  Optical network for Broadband
                                       European Leadership"                                e1686d81-2994-497c-93f6-
                                                                                    b2e913b07f59.docD06_DRAFT_V1
                                                                                                6JW2 Andreas zweiter
                                                                                         Teil.docD06_DRAFT_V15ejw




[P807]          Project P807, S. Bonaventura, et. All. - JUPITER2 – “Joint Usability, Performability
                and Interoperability Trials in Europe” http://www.eurescom.de/~public-webspace/P800-
                series/P807/results/Acceptability/R2/D2-T4-Acceptability-R2-RealisticTasks.pdf

[Rosenbaum] G. Rosenbaum, W. Lau, and S. Jha, “Recent directions in virtual private network
            solutions”, In proceedings of ICON 2003, pages 217-223, Sydney, September
            2003
[Schulz]        H. Schulzrinne, A. Rao, R. Lanphier “Real Time Streaming Protocol (RTSP)”, IETF
                RFC2326, April 1998.
[Roberts1]      J. W. Roberts, “Multiservice traffic performance in mobile networks”, available at
                http://www-sop.inria.fr/mistral/personnel/ K.Avrachenkov/WiOpt/PDFfiles/RobertsInv.pdf
[Chu]           J. Chu, K. Labonte, and B. N. Levine, "Availability and locality measurements of
                peer-to-peer file sharing systems", in Proc. of SPIE ITCom: Scalability and Traffic
                Control in IP Networks, vol. 4868, July 2002.
[Gummadi]       . K. P. Gummadi, R. J. Dunn, S. Saroiu, S. D. Gribble, H. M. Levy, J. Zahorjan,
                “Measurement, Modeling, and Analysis of a Peer-to-Peer File-Sharing Workload”,
                Proceedings of the 19th ACM Symposium on Operating Systems Principles
                (SOSP-19), October 2003.
[Lievens]       I. Lievens et al. “Contribution to A2.2.2 – Traffic measurement, characterization,
                and modeling”, NOBEL Internal Report.
[Shenker]       S. Shenker, "Fundamental Design Issues for the Future Internet", September
                1995, IEEE Journal On Selected Areas In Communications, Vol.13, No.7
[Roberts2]      J. W. Roberts, “Multiservice traffic performance in mobile networks”, available at
                http://www-sop.inria.fr/mistral/personnel/ K.Avrachenkov/WiOpt/PDFfiles/RobertsInv.pdf
[Discman]       EURESCOM project Discman, "Service Models and Architectures", EDIN 0063-
                1006
[Andrew]        Andrew Lord et al., "Definition and Description of Physical Layer Functions for
                Transparent Optical Networks", NobelNOBEL deliverable report WP5-D003, May
                2004.
[McGuire]       Alan McGuire, Shehzad Mirza, and Darren Freeland, "Application of Control Plane
                Technology to Dynamic Configuration Management", IEEE Communications
                Magazine, pp. 94-99, September 2001.
[ASON]          ITU-T Rec. G.8080/Y.1304, "Architecture for the Automatic Switched Optical
                Networks (ASON)", November 2001.




         1.2.2 Acronyms
AAA          Aauthentications, Aauthorization and Aaccounting
AP           Access Point
ASON         Automatic Switched Optical Network




                                               Page 9 of 168
                                IST IP NOBEL "Next generation
                                                                                Title of the document
                                Optical network for Broadband
                                     European Leadership"                 e1686d81-2994-497c-93f6-
                                                                   b2e913b07f59.docD06_DRAFT_V1
                                                                               6JW2 Andreas zweiter
                                                                        Teil.docD06_DRAFT_V15ejw




ASTN        Automatic Switched Transport Network
ATM         Asynchronous Transfer Mode
BER         Bit Error Ratio
BGP         Border Gateway Protocol
CapEx       Capital Expenditure
C-band      Optical vacuum wavelength range 1530-1565 nm
CBR         Constant Bit Rate
CE          Customer edge, customer equipment
CL          Connection Less
CO          Connection Oriented
CoS         Class of Service
CPE         Customer premisis equipment
DCD         Digital Cinema Distribution
EDFA        Erbium doped fibre amplifier
EML         Element Management Layer
E-NNI       External Network-Network Interface (both inter-carrier and intra-carrier)
ESCON       Enterprise Systems Connection
FC          Fiber Channel
FCAPS       Fault, Configuration, Accounting, Performance, and Security Management
FEC         Forward error correction
FICON       Fibre Connectivity
FR          Frame Relay
FTP         File Transfer Protocol
GDPF        Geographical Dispersed Parallell Sysplex
GMPLS       Generalized Multi Protocol Label Switching
GPID        Generalized payload identity
GRE         Generic Routing Encapsulation
Hyperfine   WDM with very narrow-band, densely spaced wavelength channels
WDM
IDS         Intrusionde Ddetection Ssystems
IP          Internet Protocol
IPSec       Secure Internet Protocol (IETF)
IPv4,       Internet Protocol version 4, and 6
IPv6



                                        Page 10 of 168
                          IST IP NOBEL "Next generation
                                                                       Title of the document
                          Optical network for Broadband
                               European Leadership"               e1686d81-2994-497c-93f6-
                                                           b2e913b07f59.docD06_DRAFT_V1
                                                                       6JW2 Andreas zweiter
                                                                Teil.docD06_DRAFT_V15ejw




IP-TV   Internet Protocol Television
I-SAN   Inter-SAN
ISP     Internet Service Provider
ITU-T   International Telecommunication Union Telecommunication standardisation section
L1      Transport Layer 1
L2      Transport Layer 2
L2SC    Layer 2 Switching Capability
L3      Transport Layer 3
LAN     Local Area Network
LDP     Label Distribution Protocol
LSP     Label Switched Path
MAC     Media Access Control
MP2MP   Multi Point to Multi Point
MP2P    Multi Point to Point
NML     Network Management Layer
MP      Management plane
MPLS    Multi Protocol Label Switching
NNI     Network-Network Interface
NSP     Network Service Provider
OADM    Optical Add/Drop Multiplexer
OBS     Optical burst switching
ODU     Optical channel Data Unit
OCDMA   Optical Code Division Multiple Access
OCH     Optical channel
OEO     Optical to Electrical to Optical conversion
OEXC    Optoelectric cross-connect
OpEx    Operational Expenditure
OPM     Optical performance monitoring
OPS     Optical packet switching
OSNR    Optical Signal to Noise Ratio
OTN     Optical transport network
P2MP    Point-to-point services
P2P     Peer-to-Peer




                                      Page 11 of 168
                                IST IP NOBEL "Next generation
                                                                              Title of the document
                                Optical network for Broadband
                                     European Leadership"                 e1686d81-2994-497c-93f6-
                                                                   b2e913b07f59.docD06_DRAFT_V1
                                                                               6JW2 Andreas zweiter
                                                                        Teil.docD06_DRAFT_V15ejw




PDU         Protocol Data Unit
PE          Provider edge, provider equipment
PMD         Polarisation mode dispersion
PoP         Point of Presence
PSN         Circuit Emulation Services over Packet Switched Network
PSN         Packet Switched Network
PWE3        Pseudo Wire Emulation Edge-to-Edge
QoS         Quality of Service
RD          Route Distinguisher
RSVP-TE     Resource Reservation Protocol-Traffic Engineering (MPLS)
RTSP        Real Time Streaming Protocol
SAN         Storage Area Network
SC          Switching Capability
SLA         Sercive Level Aggrement
SRDF        Symmetrix Remote Data Facility
SoD         Storage on Demand
SOHO        Small Office / Home Office
TCP         Termination Connection Point
TDM         Time Division Multiplexing
TFP         Termination Flow Point
TG          Transit Group
TMF         Telemanagement Forum
VPN         Virtual Private Network
VBR         Variable Bit Rate
VLBI        Very Low Baselin Interferometry
VPN         Virtual Private Network
WAN         Wide Area Network
WP          Working Package
UNI         User to Network Interface
XML         Extensible Markup Language


 1.3      Document History
Version    Date          Authors                                Comment




                                        Page 12 of 168
                              IST IP NOBEL "Next generation
                                                                                   Title of the document
                              Optical network for Broadband
                                   European Leadership"                     e1686d81-2994-497c-93f6-
                                                                     b2e913b07f59.docD06_DRAFT_V1
                                                                                 6JW2 Andreas zweiter
                                                                          Teil.docD06_DRAFT_V15ejw




0.01      01/03/2004     Antonio Manzalini,                     Initial document template and TOC
                         F.-Joachim Westphal
0.02      11/05/2004     Antonio Manzalini, Andrea Di Giglio,   First drafts of Tilab, Telia, Acreo and
                         Anders Berntson, Antonio J.            Telefonica contributions
                         Elizondo
0.71      22/06/2004     Björn Hermansson                       Updated draft
                         F.-Joachim Westphal
0.8       15/07/2004     F.-Joachim Westphal                    Updated draft
                                                                First draft chapter 5 included
0.9       06/08/2004     F.-Joachim Westphal                    Update of chapter 5 and 6
1.0       11/08/2004     Hans Carlden,                          Update of chapter 4,
                         F.-Joachim Westphal
                                                                Simplification of structure of chapter 6
1.4       11/09/2004     All partners                           Update and harmonization of all
                                                                chapters
14        13/09/2004     Andrea de Gigilo,                      Extented Management Summary
                         Antonio Manzalini,                     included; Update of chapter 3,
                         F.-Joachim Westphal                    Questionnaire included
15        17/09/2004     Anders Berntson,                       Final Editing
                         Andrea de Gigilo,
                         Antonio Manzalini,
                         Andreas Gladisch,
                         F.-Joachim Westphal



 1.4     Document overview
The number and scale of telecommunications-based systems using broadband capabilities
is increasing, and they are becoming more and more common in everyday life. There are a
set of underlying network technologies and services that are required for all of these
systems, and the technologies and services are the topic of increasing interest. The key to
project success in the current, constantly evolving tTelecommunications environment is to
gather as much information and high-level requirements to those systems as possible and
utilize them as effectively as possible.
To achieve these aims, particular attention is being paid in this document to how the
utilization of requirements to emerging application can be directly linked to relevant network
service definition activities.




. Furthermore, we investigate the generic traffic typology and undertake the derivation and
positioning of the collected application requirements and service alternatives for integration
into a framework of future networks that covers architecture, network requirements, and
service requirements, all suitable for the diverse range of applications, and have yet to be
perfected.



                                        Page 13 of 168
                              IST IP NOBEL "Next generation
                                                                                 Title of the document
                              Optical network for Broadband
                                   European Leadership"                    e1686d81-2994-497c-93f6-
                                                                    b2e913b07f59.docD06_DRAFT_V1
                                                                                6JW2 Andreas zweiter
                                                                         Teil.docD06_DRAFT_V15ejw




However, at present the requirements to on network control and management
technologies, derived utilizingutilising the diverse range of application and service
information described in this document, are still incomplete. The same reflection is true for
                                                                                                     Formatted: Font: Not Bold, English (United
network infrastructure, node architecture, shortly for all the activities in the project.,           Kingdom)
                                                                                                     Formatted




. Therefore, this document attempts also to iron out in a generic way what these application
and service requirements may bring to these diverse activities with a first list of detailed
requirements presented at the end of the document.




   Table 111: Requirement chain and top-down approach adopted in NobelNOBEL




                                       Page 14 of 168
                              IST IP NOBEL "Next generation
                                                                                  Title of the document
                              Optical network for Broadband
                                   European Leadership"                     e1686d81-2994-497c-93f6-
                                                                     b2e913b07f59.docD06_DRAFT_V1
                                                                                 6JW2 Andreas zweiter
                                                                          Teil.docD06_DRAFT_V15ejw




 2        Extended Management Summary

 2.1      Definition of the Context
In the last decade Network Providers largely deployed Sonet/SDH technology in core and
metro transport networks. Furthermore, investments have been made in the last few years
for introducing WDM point-to-point systems in the core long-haul networks.
One of the major overall problems for the evolution of current transport networks seems to
be that there are not enough demands (killer applications) to use/deploy more capacity and
vice-versa that costs are currently too high to fuel a large adoption of new broadband
applications/services (as new applications/services are expected to replace legacy ones
improving the quality perceived by the Customers). In this context, the evolution of
transport networks is likely to be lead by a few elementary drivers, such as: new network
solutions have to
     -   optimise the use of resources (reducing CAPEX);                                              Formatted: Bullets and Numbering

     -   have to reduce the operating costs (reducing OPEX); have to
     -   improve quality, efficiency in providing current and new services (increasing and
         generating new revenues).
The architectural studies of NOBEL (within Work Package 1) are aimed at defining high-
level requirements (technical and business oriented), technologies and solutions for the
evolution of core and metro networks. The access segment (already covered by the IST IP
Muse),) will be considered only from the perspective of the aggregated traffic offered to the
metro and the core networks. Nevertheless a cooperation with IST IP Muse is already
established in order to define an overall architecture for providing end-to-end broadband
services.
Specifically, from a technology viewpoint, NOBEL is mainly focussing on ASON/GMPLS
architectures as they seem to be the most widely recognised solution to match the
emerging requirements, whilst allowing different levels of integration of network
functionality (e.g. L3 routing, and L1-L2 switching) and more flexible control / management
strategies.
Basically the NobelNOBEL network is providing three main classes of network services
(e.g. Virtual Privat Networks at layerL 1, layer L2 and layer L3). Different services can be
provided by separate networks, or different services may share same network resources.
Figure 1 shows two examples of these service networks; in particular, the upper scenario
represents a generic storage service. The CPE is represented by the data-machine
interface, the internal Storage Area Network collects the traffic coming from the CPEs and
it is the fFeeder (metro hub) depicted in Figure 1; in this scenario a layer 1 network service
is imagined, so there are optical equipments (OADMS) between the metro and the core
network.
The scenario depicted in the lower section of Figure 1 represents a typical residential
Internet application where the Personal Computer / Modem is represents the CPE. Inside
the ISP (Internet Service Provider) building it is possible to find the collector of residential



                                       Page 15 of 168
                                         IST IP NOBEL "Next generation
                                                                                                     Title of the document
                                         Optical network for Broadband
                                              European Leadership"                         e1686d81-2994-497c-93f6-
                                                                                    b2e913b07f59.docD06_DRAFT_V1
                                                                                                6JW2 Andreas zweiter
                                                                                         Teil.docD06_DRAFT_V15ejw




traffic (fFeeder) and, from the downstream side, the point of injection toward of the
collected traffic towards the metro network (PoP metro).




                             Nobel

          Server             gateway/                                                 gateway/
                             extender                                                 extender


                                      metro                  core                 metro
                                                                           ODXC                             Storage
                                              ODXC                                               Storage
                   Storage                                                                                 Subsystem
      Storage       Area                                                                          Area
     Subsystem     Network                                                                       Network
                             metro PoP




 Customer’s Data Center                                                                     SSPr’s Data Center
                                               core PoP
                              Feeder




    CPE
                                ISP




                 access                  metro                   core



                             Nobel
Figure 111: Examples of network services for storage and residential applications


An essential starting point of the discussion of the NOBEL vision fFrom a general functional
perspective, the NobelNOBEL vision is in line withis the ITU Recommendation Y.2011,
which introduces two strata to describe the network: Services and Transport Stratum
(Figure 2Figure 2Figure 2).




                                                          Page 16 of 168
                              IST IP NOBEL "Next generation
                                                                                        Title of the document
                              Optical network for Broadband
                                   European Leadership"                             e1686d81-2994-497c-93f6-
                                                                             b2e913b07f59.docD06_DRAFT_V1
                                                                                         6JW2 Andreas zweiter
                                                                                  Teil.docD06_DRAFT_V15ejw




                                                          Management Plane
                                                           Control Plane
                                                          User Plane


                                     NGN Service Layer


                                                          Management Plane
                                                           Control Plane
                                                          User Plane


                                    NGN Transport Layer


                   Figure 222: Functional perspective of the network


The transport stratum transports digital information, it interconnects users and/or service
nodes. Specifically NobelNOBEL intends to generalised the transport stratum to a
resources stratum (including storage, computing, sensors and other H/W, S/W distributed
resources).
The Services stratum provides the users service, it uses the transport layer for building the
services (e.g. connectivity).
The transport stratum can be further subdividedlayered based on the layering principles as
defined in ITU G.805/G.809. For example the transport network can be realized using ETH,
SDH and OTN layers.
From the NOBEL perspective For the services stratum two kindtwo kinds of services can
be defined for the services stratum:
      Applications (i.e. User services): These services can be used by humansHumans
       can use these services. Examples are web services, telephony, video conferencing,
       videoand video on demand.
      Network services :services: These services provide connectivity to the user, for
       example a Layer 2 VPN. This connectivity is used by user services, like telephony.


 2.2     Emerging Applications

       2.2.1 Definitions and goal of the research
The aim of this section is to analyze the characteristics of the most important client (user)
applications and to , having the goal of establishing a mappingcorrespondence between
applications and the network services that might be the most suitable for carrying them.
It is important to note how really is an user application and which is the difference
distinguish between applications and network services; basically an user application is
defined as a source of traffic from client (user) side; so, for instance, a leased line stand
alone is not an user application, instead disck back-up, that uses a leased line for transport
its data, is an user application.
The analyzed emerging user applications are grouped into three main categories:




                                       Page 17 of 168
                              IST IP NOBEL "Next generation
                                                                                Title of the document
                              Optical network for Broadband
                                   European Leadership"                   e1686d81-2994-497c-93f6-
                                                                   b2e913b07f59.docD06_DRAFT_V1
                                                                               6JW2 Andreas zweiter
                                                                        Teil.docD06_DRAFT_V15ejw




      Storage;
      Grid Ccomputing;
      Multimedia.
The first two categories (storage and gGrid computing) are expected to generate high
levels of traffic volume, and are strongly related to business customers. Presently the
market penetration of this applications isbut a relatively small penetration into the market,
but this market segment is continueally growing., Tthe last category (multimedia
applications) probably generate a level of traffic (traffic volume or bandwidth) lower than
sStorage or Grid, but, onfrom the other side, it is expected that it will have a granular
penetration, ever since the applications consisting of this group are addressesd toward
SOHOs and residential customers.
It is not the focus of this study to giveThis research is not focused on giving an estimate of
the dimension of the network needed to transport the bandwidth generated by the reported
applications.


       2.2.2 Parameters characterizing emerging applications
The research was leaded made from a more technological perspectivepoint of view;; in
fact, for each user application, the effort was concentrated to valuate some parameters
about the parameters were grouped into the following main classes for characterizing the
applications.
      Resilience (information about the survivability of an existing connection);
      Provisioning (information about the rate of set-up or tear down a connection);
      Dynamics (information about if/which the parameters of an existing connection that
       can be modified without tear-down, for instance connectivity, QoS parameters
       and/or bandwidth);
      Bandwidth (information about the amount of peak and average bandwidth and its
       granularity);
      QoS (information about whole latency, and differential latency variation,time and
       data integrity on an existing connection);
      Connectivity (connection topology);
      Transparency (Tthe primary framing protocols that can be carried on the
       connection);
      Interface (information about physical interface requirement).
More in particular a table, simplified compared to the one presented in to the document,
was produced. The conclusions were summarized in a table, which can be found in chapter
4. In this paragraph a simplified version of the results is presented is presented in Table 2.




                                       Page 18 of 168
                                                    IST IP NOBEL "Next generation
                                                                                                                                                                                                                                              Title of the document
                                                    Optical network for Broadband
                                                         European Leadership"                                                                                                                                      e1686d81-2994-497c-93f6-
                                                                                                                                                                                                            b2e913b07f59.docD06_DRAFT_V1
                                                                                                                                                                                                                        6JW2 Andreas zweiter
                                                                                                                                                                                                                 Teil.docD06_DRAFT_V15ejw




            User Application             Resilience      Provisioning Dynamics
                                                                     Ma                                                   Bitrate [Mb/s]                                                     QoS                              Connectivity   Transparency        Interface
                                                                     x
                                                                     set
                                                                     -                                                            Ma                           Mi
                                                                     up/                                                          x                            n
                                                                     tea                                                          No                           Dy
                                                    Mi
                                                                     r-                                                           mi                           na                                           Da
                                                    n
                                                                     do                                                           nal                          mi                                           ta
                                                    av
                                                                     wn                                                           Bitr                         c                         Lat                int
                                                    ail
                                                                     tim                                                          ate                          ran                       en                 egr
                                                    abi
                                                    lity             e                                                                                         ge                        cy                 ity



Storage
- Back-Up / Restore                      Gold                          min                                    None               400 const                                         DEL             Gold                      p2p             L1-L2-L3       FC/IP/GbE
- Storage on demand (SoD)                Platinum                      sec                                    BR & QoS          1000                                               HRT             Platinum                  p2p             L1-L2-L3       FC/IP/GbE
- Asyncrhonous mirroring                 Platinum                      sec                                    BR & QoS           400                                               RT              Gold                      p2p             L3             IP/GbE
- Synchronous mirroring                  Platinum                      min                                    None              2000 const                                         HRT             Platinum                  p2p             L1             FC/WDM
Grid computing
- Compute Grid                           Gold                          sec                                    Full       N.A.                           N.A.                       RT              Platinum N:M                              L3             IP
- Data Grid                              Gold                          sec                                    Full       N.A.                           N.A.                       RT              Platinum N:M                              L3             IP
- Utility Grid                           Platinum                      sec                                    Full       N.A.                           N.A.                       RT              Platinum N:M                              L3             IP
Multimedia
- Video on Demand                        Silver                        sec                                    None                5 const                                          DEL             Gold                      p2p             L3             IP
- Video Broadcast (IP-TV)                Silver                        sec                                    None                5 const                                          DEL             Gold                      multi-broad     L3             IP
- Video Download                         Gold                          sec                                    None               20      19                                        DEL             Gold                      p2p             L3             IP
- Video Chat                             Silver                        sec                                    None        0.2-0,4       0,5                                        RT              Silver                    p2p             L3             IP
- Narrowband Voice, data (VoIP,...)      Platinum                      ms                                     None       5,3-64 kb N.A                                             RT              Gold                      p2p             L3             IP
- Telemedicine (disgnostic)              Platinum                      ms                                     None       1-10       N.A.                                           RT              Gold                      p2p             L1-L2-L3       IP/GbE
- Gaming                                 Silver                        sec                                    None       10-25 kb       0,5                                        RT              Silver                    p2p             L3             IP
- Digital distribution, digital cinema   Gold                          sec                                    None            1000 const                                           RT              Gold                      p2p-multi       L3             IP
- Video conference                       Gold                                                                 None        1,5-3,0 const                                            RT              Silver                    p2p-multi       L3             IP/GbE

            User Application             Resilience                    Provisioning                           Dynamics     Bitrate [Mb/s]                                                     QoS                             Connectivity Transparency          Interface
                                                                                  Max set-up/tear-down time




                                                                                                                                  Max Nominal Bitrate


                                                                                                                                                               Min Dynamic range
                                                    Min availability




                                                                                                                                                                                                            Data integrity
                                                                                                                                                                                         Latency




Storage
- Back-Up / Restore                      Gold                          min                                    None               400 const                                         DEL             Gold                      p2p             L1-L2-L3       FC/IP/GbE
- Storage on demand (SoD)                Platinum                      sec                                    BR & QoS          1000                                               HRT             Platinum                  p2p             L1-L2-L3       FC/IP/GbE
- Asyncrhonous mirroring                 Platinum                      sec                                    BR & QoS           400                                               RT              Gold                      p2p             L3             IP/GbE
- Synchronous mirroring                  Platinum                      min                                    None              2000 const                                         HRT             Platinum                  p2p             L1             FC/WDM
Grid computing
- Compute Grid                           Gold                          sec                                    Full       N.A.                           N.A.                       RT              Platinum N:M                              L3             IP
- Data Grid                              Gold                          sec                                    Full       N.A.                           N.A.                       RT              Platinum N:M                              L3             IP
- Utility Grid                           Platinum                      sec                                    Full       N.A.                           N.A.                       RT              Platinum N:M                              L3             IP
Multimedia
- Video on Demand                        Silver                        sec                                    None                5                     const                      DEL             Gold                      p2p             L3             IP
- Video Broadcast (IP-TV)                Silver                        sec                                    None                5                     const                      DEL             Gold                      multi-broad     L3             IP
- Video Download                         Gold                          sec                                    None               20                          19                    DEL             Gold                      p2p             L3             IP
- Video Chat                             Silver                        sec                                    None            0,64                          0,5                    RT              Silver                    p2p             L3             IP
- Narrowband Voice, data (VoIP,...)      Platinum                      ms                                     None       5,3-64 kb                      N.A                        RT              Gold                      p2p             L3             IP
- Telemedicine (disgnostic)              Platinum                      ms                                     None       1-10                           N.A.                       RT              Gold                      p2p             L1-L2-L3       IP/GbE
- Gaming                                 Silver                        sec                                    None       10-25 kb                           0,5                    RT              Silver                    p2p             L3             IP
- Digital distribution, digital cinema   Gold                          sec                                    None            1000                      const                      RT              Gold                      p2p-multi       L3             IP
- Video conference                       Gold                                                                 None              0,5                     const                      RT              Silver                    p2p-multi       L3             IP/GbE


Table 222: Characterization of applications


The effort to fill this table was concentrated to group several parameters regarding a
macro-item (i.e. Resilience, QoS, Dynamics,…) into one or at least two more qualitative
items that can help to give an idea on what kind of requirements one single user
application need and so to help to understand which network services might preferably



                                                                                 Page 19 of 168
                              IST IP NOBEL "Next generation
                                                                                    Title of the document
                              Optical network for Broadband
                                   European Leadership"                     e1686d81-2994-497c-93f6-
                                                                     b2e913b07f59.docD06_DRAFT_V1
                                                                                 6JW2 Andreas zweiter
                                                                          Teil.docD06_DRAFT_V15ejw




carry these applications. In order to give a qualitative idea of the requirements of a single
user, some sets of parameter values suitable for a group of applications were defined, (e.g.
QoS requirements for real-time applications).
At this point it is necessary to explicit explain some terms present in the table
      Resilience: three classes of availability: "Platinum" (mean avail. >99.999%, max
       recovery time < 50 ms), "Gold" (mean avail. >99.99%, max recovery time <200ms),
       "Silver" (less severe requirements about availability and max recovery time);
      Dynamics        : the parameters of an existing connection that the client can modify
       are: bandwidth (bit-rate), QoS parameters (latency, availability, data integrity) and
       connectivity (end-points of the connection). It is not very significant to have the
       possibility to modify the connectivity without having the possibility to change bit-rate
       and QoS, so three items about dynamicity are allowed: "None" (it is not possible to
       modify any parameters of an existing connection), "Bit-rate and QoS" and "Full"
       (Bit-rate, QoS parameters and connectivity modifications are allowed);
      QoS – latency: analyzing the data drawn up from the research it is possible to note
       that the maximum difference of latency is about 10% of the max whole latency; so
       the idea is to cluster the parameters about latency into only one item: "HRT" (Hard
       Real Time: Latency <10ms, Max diff. lat. <1ms); "RT" (Real Time: Latency <
       500ms, Max diff. lat. < 50-100ms); "DELAYED" (less severe requirements about
       latency); real-time video may be left out of this classification, ever since they have
       no problem about latency, but they need a severe requirement for the difference of
       latency (very lower than 10% of the whole latency allowed); this problem may be
       solved admitting a less severe requirement about the difference of latency on
       condition that a buffer will be introduced upstream the final user.
      QoS - data integrity: three classes of QoS data integrity: "Platinum" (Packet Loss
       between 0.1 and 0.5% or BER<10-12); "Gold" (Packet Loss between 0.5 and 1%);
       "Silver" (Packet Loss > 1%).


       2.2.3 Conclusions
The effort of giving a characterization of the emerging user applications leadsed to
interesting considerations.
      MostThe most part of emerging applications requires a high level of QoS and in
       particular an excellent level of availability. In effect, the decreasing cost of
       bandwidth, the inclinationtrend of the consumers to spend more money for network-
       based multimedia applications network-based, the competition between emerging
       multimedia applications (i.e. IP-TV, video-conference,…)conference…) with
       incumbent services atwith a very high level of availability and QoS (PSTN, satellite
       and cable TV,…)TV…) are factors that move the focus toward the increase of
       availability and to give a remarkablehuge differentiation about QoS for the different
       applications;
      The rising popularity of new applications such as real-time multimedia and peer-to-
       peer networking, the increasing of penetration of network-based storage
       applications network-based lead to a more dynamic situation; this is
       testifiedmanifested in by the severe requirements abouton the maximum time



                                        Page 20 of 168
                               IST IP NOBEL "Next generation
                                                                                  Title of the document
                               Optical network for Broadband
                                    European Leadership"                    e1686d81-2994-497c-93f6-
                                                                     b2e913b07f59.docD06_DRAFT_V1
                                                                                 6JW2 Andreas zweiter
                                                                          Teil.docD06_DRAFT_V15ejw




       allowed to set-up and tear–down the application (in one word “fast-provisioning”).
       Ever Ssince the majority of applications need a large quantity of bandwidth, the
       solutions to face of the problem of supply fast provisioning are essentially two:
           o   by a very (expensive) huge over-provisioning on the physical layer (currently
               service providers need to design for peek traffic volumes) for carrying traffic
               and a large amount of free bandwidth for protection/restoration necessary to
               keep the desired level of availability;
           o   to adopt the ASON/GMPLS technologies that allow the possibility of having
               dynamic connections on physical layer that might be used for bandwidth
               requests both for carrying the applications’ data and for activate only on-
               fault some restoration resources.


 2.3     Traffic Prediction and Characterizations

       2.3.1 Traffic typology and relations concerning core and
             metro networks
Traffic typology deals with the study and analysis of the traffic based on traffic types or
traffic categories. The objective of this chapter is not only to identify the main traffic types
that are to be found flowing through core or metro networks, but also to infer the impact
that these different types of traffic have on the architectural design of the core and metro
networks.
The traffic categorization approach used in this document has a deductive character,
instead of an empirical approach. It will be mainly based on criteria that strongly depend on
the given services, as well as on their implementation and their social acceptance and
success.


       2.3.2 Criteria for traffic classification
      Content orientation: a classification of traffic generated by the different
       applications could be based on the following criteria:
           o   Elasticity
           o   Interactivity level
           o   Traffic Asymmetry
           o   Service Availability
           o   Bandwidth
           o   Traffic variability
       The importance of these criteria is relative, and depends on the considered point of
       view. From the applications point of view, QoS is one of the factors to be
       considered. Thence, elasticity and levels of interactivity levels are the parameters
       that have a more clear impact on QoS.




                                       Page 21 of 168
                               IST IP NOBEL "Next generation
                                                                                     Title of the document
                               Optical network for Broadband
                                    European Leadership"                      e1686d81-2994-497c-93f6-
                                                                       b2e913b07f59.docD06_DRAFT_V1
                                                                                   6JW2 Andreas zweiter
                                                                            Teil.docD06_DRAFT_V15ejw




      Topological and geographical distribution (these topics are specially broached
       from a core/metro network point of view)
           o   Traffic generation dynamics
           o   Traffic sources scattering
           o   Local properties of traffic
       Any combination of the previous attributes can be present in the different types of
       traffic. Due to this fact, instead of inspecting every type of traffic, the identification of
       the dominant applications in a given traffic mix can be used to define the most
       relevant traffic types that would be:
                Residential:
               o   Media distribution (video, radio)
               o   P2P, e-mail (background traffic)
               o   Web-browsing
               o   VoIP, interactive gaming
                Institutions:
               o   Intra-institution (VPN, leased lines, SAN)
               o   Inter-institution (Web-browsing, web-hosting, e-business)
               o   Institution-to-customers (web-hosting, on-line shopping)


       2.3.3 Traffic classes
Although there are many criteria to classify traffic, as has been indicated previously, it is
useful to do it in function of the QoS requirements, as different applications show
completely different sensitivity to the QoS.
The most relevant criteria for QoS purposes isare the inelastic/elastic classification. In
order to produce a more detailed classification, a further criterion can be introduced
(interactive and non-interactive), which produces the following four classes of traffic:
                                          Elastic        Inelastic
                    Interactive      Elastic interactive Real Time
                                      (Transactional) (Conversational)

                    Non                  Elastic non           Streaming
                    interactive          interactive


                   Table 333: Traffic Types




                                        Page 22 of 168
                              IST IP NOBEL "Next generation
                                                                                   Title of the document
                              Optical network for Broadband
                                   European Leadership"                     e1686d81-2994-497c-93f6-
                                                                     b2e913b07f59.docD06_DRAFT_V1
                                                                                 6JW2 Andreas zweiter
                                                                          Teil.docD06_DRAFT_V15ejw




       2.3.4 Traffic characteristics per application and per user
The different characteristics of traffic should be studied, making a characterization of both
the applications and the users, in order to be able to make traffic forecasts, network
provisioning and define the QoS classes needed in a given network, as the characteristics
of traffic impact on the core and metro networks architecture.
As an example, tThe main traffic types over Anella Científica (the Catalan R&D Network)
are web traffic and P2P traffic, although incoming and outgoing traffic show different
characteristics. The weekly traffic of the same network show that week days and weekend
days have a completely different profile that should be separately studied and taken into
consideration when making traffic characterization. Another significant characteristic is the
different types of traffic present different asymmetries. Different applications have different
packet sizes, which may help to identify them and also to identify disguised P2P traffic.
The level of traffic aggregation in core networks produces different traffic profiles
whichprofiles that have to be taken into consideration.
NOBEL WP2 will also provide users profiles for different types of users. During the
labourworking day for a typical residential broadband user, aggregated upstream traffic is
very uniform during the whole labourworking day, whereas aggregated downstream traffic
doubles up during labour hours. Traffic is becoming more and more symmetric.


       2.3.5 Impact of traffic on the network architecture

                       2.3.5.1 Network switching paradigms
The two different network switchingnetwork-switching paradigms for transmitting messages
through a telecommunication network are circuit switching and packet switching. At least
for some years, optical networks will continue being circuit oriented and therefore more
suitable for conveying inelastic traffic, although of course, perfectly able to transport elastic
traffic. As core networks will normally be optical based, circuit switchingcircuit-switching
paradigm will prevail in these networks. However, in metro networks a higher range of
technical possibilities can be deployed. Therefore, metro networks can more easily be
packet switching based.

                       2.3.5.2 Topology of the network
When designing the network, a given topology must be chosen. As with connection
dynamics, should a metro or core network be mostly devoted to media distribution a star or
tree orientedtree-oriented topology could be possibly the best solution, as it would minimise
the amount of deployed resources.
However, and due in part due to the preponderance that P2P traffic is getting, and in part
due to the need for to being able to satisfy high and variable traffic demands with an
optimised solution, is expected that current metro and core networks achieve a partially (if
no fully) meshed topology. There It is also important to take into account that the traffic mix
in core and metro networks can be different.




                                        Page 23 of 168
                            IST IP NOBEL "Next generation
                                                                             Title of the document
                            Optical network for Broadband
                                 European Leadership"                  e1686d81-2994-497c-93f6-
                                                                b2e913b07f59.docD06_DRAFT_V1
                                                                            6JW2 Andreas zweiter
                                                                     Teil.docD06_DRAFT_V15ejw




                     2.3.5.3 Connection dynamics
In a dynamic connection oriented scenario, a complex control plane (CP) and network
management (NM) facilities are required. A core or metro network should be mainly
devoted to media distribution, this network could be easily optimised with manually
establishment mechanisms. Same can be applied when traffic aggregates that flow in the
network are well characterised and are rather constant.
Notwithstanding, core networks are expected to be more dynamic than current ones in
order to be able to provide a set of complex and different services on demand, being able
not only to serve to big customers (ISPs, ASPs, RSPs, big enterprises) but also even to
medium-size customers (SOHOs). Metro networks are expected to be much more dynamic
than core networks, as on demand services will be provided with less granularity than in
core networks. This connection dynamics will also increase when delivering special
services as on demand VPN connectivity.


 2.4    Network Modes and Services

       2.4.1 Network Modes
There are 2 main network modes: circuit-switched and packet switched; packet-switched
networks can be connectionless or connection-oriented.
In circuit-switched layer networks, a path based on a physical link, optical wavelength or
TDM timeslot is established and dedicated to a single connection between APs in the
network for the duration of the connection.
In packet-switched networks, packets are normally forwarded based on information in the
packet header. Packet switching provides connectivity while making efficient use of
network resources by sharing them with many users (based on the assumption that not all
users need to use the resource all of the time). Packet-switched networks can be
connectionless or connection-oriented.
In connectionless packet-switched networks, once the data is sent, the connection is
broken until further information is either sent or received.
In connection-oriented packet-switched layer networks, connections are established and
maintained until connectivity is no longer required (regardless of whether data is been
transmitted or not).




                                     Page 24 of 168
                               IST IP NOBEL "Next generation
                                                                                 Title of the document
                               Optical network for Broadband
                                    European Leadership"                     e1686d81-2994-497c-93f6-
                                                                      b2e913b07f59.docD06_DRAFT_V1
                                                                                  6JW2 Andreas zweiter
                                                                           Teil.docD06_DRAFT_V15ejw




                               Layers   L1          L2           L3

               Network modes
               Connection- oriented                 ATM        MPLS
               Packet-switched
               Connection-oriented      VC (SDH)
               Circuit-switched         ODU (OTH)
                                        OCh (OTN)

               Connectionless                       Ethernet     IP
               Packet-switched




                        Table 444: Network modes vsvs. Layers


       2.4.2 Network Services
This section on Network Services describes threefive assigned groups of network services:
their features, relations, and differences. The five groups are: Public IP, Business IP, and
Virtual Private Networks (VPN) on transport layer 3, VPN on layer 2, and VPN on layer 1.
As additional network services based on Layer 3 Public IP, Business IP could be assumed.
The characteristics of each network servicegroup are described along with their
performance parameters.
The different layers’ VPNs are explained from a perspective of connectivity, control,
scalability and flexibility. Mechanisms enabling the network “privacy” are described, such
as MPLS and tunnellingtunneling in packet networks, and lightwave services enabled by
some future L1-VPN. Finally there is considered how network services should match the
applications emphasized in chapter 4: storage, grid, and multimedia.
The information in this chapter is based on two tables. The first table presents typical data
and limiting performances for the five groups of network services. The second table deals
with the mapping of the emerging applications, defined in chapter 4, onto the network
services.




                      2.4.2.1 Network service table

Three classes of Network Services have been identified based on the related network
layers in which they are provided.




                                        Page 25 of 168
                              IST IP NOBEL "Next generation
                                                                                       Title of the document
                              Optical network for Broadband
                                   European Leadership"                        e1686d81-2994-497c-93f6-
                                                                        b2e913b07f59.docD06_DRAFT_V1
                                                                                    6JW2 Andreas zweiter
                                                                             Teil.docD06_DRAFT_V15ejw




              Network                             Description
              Service
               Layer 1        Provides a physical layer (i.e. layer 1) service between
                           customer sites belonging to the same VPN. Connections can be
                           based on physical ports, optical wavelengths or TDM timeslots.

               Layer 2         Provides at data link layer (i.e. layer 2) service between
                           customer devices belonging to the VPN. Forwarding of user
                           data packets is based on information in the packets' data link
                           layer headers, e.g. DLCI, ATM VCI/VPI, or MAC addresses

               Layer 3         Provides a network layer (i.e. layer 3) service between
                           customer devices belonging to the VPN. Forwarding of user
                           data packets based on information in the Layer 3 header, e.g.
                           IPv4 or IPv6 destination address.


                               Table 555: Network Services


The network services have been generically labeled VPN at L1, L2 and L3.
For each network service, typical performances are given. The table definetable defines
the parameters that put performance limitation on the different network services and the
applicability for the new applications. About 20 parameters are investigated such as: set up
time, capacity, latency, packet loss and availability.
The IP network service is, as already mentioned, divided into Public and Business IP
where Public IP is a “best effort” service and Business IP is a higher priority class that for
example can handle latency sensitive applications. Business IP is also presumed to offer
higher bandwidth.
The VPN services on all layers, L1, L2 and L3, are divided into a permanent by configured
network service or an on-demand service. The permanent service is totally managed by
the network service provider but the on-demand service can be controlled dynamically by
the customer. L1 and L2 VPN services are further divided into high and low availability. The
high availability network services isservices are normally produced by
protection/restoration, which offer an alternative way for the traffic.


       2.4.3 Mapping of applications to services
In the second table the three application groups from chapter 4 (“Emerging applications”)
Storage, Grid computing and Multimedia are mapped onto the network services. For each
application the requirements are checked against the network service performance.




                                       Page 26 of 168
                             IST IP NOBEL "Next generation
                                                                                                                                                                                                                                                                                                  Title of the document
                             Optical network for Broadband
                                  European Leadership"                                                                                                                                                                        e1686d81-2994-497c-93f6-
                                                                                                                                                                                                                       b2e913b07f59.docD06_DRAFT_V1
                                                                                                                                                                                                                                   6JW2 Andreas zweiter
                                                                                                                                                                                                                            Teil.docD06_DRAFT_V15ejw




                                                                                                                                                                                                                                                                                                              - Narrowband Voice, data (VoIP,...)




                                                                                                                                                                                                                                                                                                                                                                            - Digital distribution, digital cinema
                                                                                                                                                                                                                                                - Video Broadcast (IP-TV)
                                                                                         - Asyncrhonous Mirroring




                                                                                                                                                                                                                                                                                                                                                                                                                                          Tele-medicine/diagnostic
                                                                                                                    - Synchronous Mirroring
                                                                   - Storage on Demand
                                             - Back-Up / Restore




                                                                                                                                                                                                                            - Video on Demand




                                                                                                                                                                                                                                                                                                                                                                                                                     - Video conference
                                                                                                                                                                                                                                                                            - Video Download
                                                                                                                                              Grid computing
                                                                                                                                                               - Compute Grid




                                                                                                                                                                                                                                                                                               - Video Chat
                                                                                                                                                                                              - Utility Grid
                                                                                                                                                                                - Data Grid




                                                                                                                                                                                                                                                                                                                                                    - Gambling
                                                                                                                                                                                                               Multimedia




                                                                                                                                                                                                                                                                                                                                                                 - Gaming
                                   Storage




Public IP
Business IP
VPN - L3    permanent
            on-demand
VPN - L2    permanent, Hi avail
            permanent, Low avail
            on-demand, Hi avail
            on-demand, Low avail
VPN - L1    permanent, Hi avail
            permanent, Low avail
            on-demand, Hi avail
            on-demand, Low avail



Table 666: Mapping applications into network services (light blue: application will run
on
        this network service; dark blue: more efficient implementation, white: no support
        for the application)
                                                                                                                                                                                                                                                                                                                                                                                                                                                        Formatted: Bullets and Numbering
       2.4.4          Quality of Service characteristics
Quality of Service is defined [ITU-T Rec. E.800] as being the collective effect of service
performances, which determine the degree of satisfaction of a user of the service.
ITU-T Rec. G.1000 provides a framework and definitions for communications quality of
service, and G.1010 defines a model for multimedia Quality of Service (QoS) categories
from an end-user viewpoint.
                                                                                                                                                                                                                                                                                                                                                                                                                                                        Formatted: Bullets and Numbering
       2.4.52.4.4 Concepts of VPN service and VPN
             trasporttransport layers
It should be noted that the same transport layers cancould be used to provide multiple VPN
services. For example a transport layer based on SDH can be used to provide Layer 1
(e.g. TDM), Layer 2 (e.g. Ethernet) and Layer 3 (e.g. IP) VPN services.
For this reason, it is worth distinguishing VPN service and transport layers. Both VPN
transport and service layers each have their own set of connectivity inputs and outputs
known as access points (APsAPs).




                                              Page 27 of 168
                              IST IP NOBEL "Next generation
                                                                                 Title of the document
                              Optical network for Broadband
                                   European Leadership"                    e1686d81-2994-497c-93f6-
                                                                    b2e913b07f59.docD06_DRAFT_V1
                                                                                6JW2 Andreas zweiter
                                                                         Teil.docD06_DRAFT_V15ejw




When the VPN service layer and VPN transport layers are different, the VPN service layer
information must be adapted for transmission across the VPN transport layer.
Examples of adaptation functions include multiplexing, coding, rate changing, and aligning
(which may also include some from of fragmentation and sequencing if the transport layer
traffic unit is smaller than the service layer traffic unit).
Even if VPN transport layer trail/connectionless trail termination functions are VPN service
layer independent, but adaptation functions must exist for each transport-service pair
defined: adaptation functions are required between APs in the VPN transport layer and
CPs/FPs in the VPN service layer.


 2.5     High-Level Network Requirements
The goal of updating of high-level requirements for next-generation metro and core
networks aligns directly with the NOBEL project priorities as articulated in the project’s
objectives. Without these enriched requirements fewer answers to research questions
would be obtained and fewer breakthroughs would be accomplished, since they allow
deriving a set of network technologies and service deployment capabilities that are
specified in the different WPs of the project.
Chapter 7 focusing on architecture, network, and service requirements recognizes
advanced networking infrastructure as a basic enabler, and as an area that presents great
opportunities for the future empowerment of broadband service deployment technology
combining enhanced Transport plane, Control plane, and Management plane.
The structuring of requirements follows, in fact, the clustering proposed by ITU NGN with
the main objective that it is time to make a concerted effort to systematically embrace the
rush of new network capabilities. Furthermore, a general requirement class is introduced.
Three essential issues were identified in the aArchitecture requirement part and rank
ordered as to their impact on the success of NOBEL network. Besides the separation of
functional planes, importantly, the network must respect the so-called end-to-end principle,
i.e. the network must not be constrained by any application. Finally, in order to allow
seamless migration, backward compatibility requirements to existing network
infrastructures is explained.
The identified network requirements are grouped into eleven items, without ranked order.
Note that several critical issues are not addressed here because they were neither
networking nor network technology issues. However, these requirements cover all the
functional planes going from the data plane and relevant physical layer features up to the
management plane requirements.
Several general observations and conclusions may be made after analyzing these network
requirements.
The first, and perhaps most significant, observation is that today´s networks are already, or
are rapidly becoming, an inherently distributed platform. Distributed means that a network
control requirements are routinely addressed and that the resulting network depends
critically, or will in the near future in real networks, on an infrastructure that supports the
process of seamless (control plane) interoperability both from intra-carrier and inter-carrier
perspectives. This leads to more and more performing Traffic Engineering requirements,
network performance enhancements, performance optimization, and performance



                                       Page 28 of 168
                              IST IP NOBEL "Next generation
                                                                                Title of the document
                              Optical network for Broadband
                                   European Leadership"                   e1686d81-2994-497c-93f6-
                                                                   b2e913b07f59.docD06_DRAFT_V1
                                                                               6JW2 Andreas zweiter
                                                                        Teil.docD06_DRAFT_V15ejw




improvement of network protocols. Even though low-level considerations are not included
here, these protocol capabilities are addressed through multi-layer and multi-domain
technology considerations. Another very innovative requirement is provided by the vertical
integration of control plane instances operating over several layers of the data planes. A
typical example lies in the MPLS+ASON/GMPLS control plane interworking, where all the
transport layers obey to a common ASON/GMPLS instance with various auto-discovery
mechanisms.
A second observation is that when asked what sort of features are needed to support
distributed control plane-enables networks, the answer always involves a lot of network
administrative and management issues beyond just basic networking capacity. This covers
security and authentication, addressing, billing and charging, as well as accounting
requirements for transport services.
A third observation is that there is considerable commonality in the automation needed by
the various services described in the previous chapters of the deliverable. This means that
requirements converge towards defining a common network infrastructure for highly
automated services such as bandwidth on demand services. (This is apparent in the tables
in each of the application sections.)
Fourth, a large part of the requirements suggests the need for high-speed networks to
couple, and manage the widely distributed, high-performance applications and services
using an accurate intra-domain provisioning mechanism that satisfies the contracts with
customers while optimising the use of the network resources. In this context, the
requirements propose that the emerging policy-based management paradigm is the
adequate means to achieve this goal.
Requirements that can be combined in order to represent more realistic networking
systems include also service requirements that are grouped into two categories: general
and QoS requirements. With reference to NGN work, it clearly appears that bandwidth
requirements are not the only measure of needed QoS performance. End-to-end transport
services with QoS in hybrid/transparent networks using cheap advanced optical
transmission are in the heart of the NOBEL network requirements per domain and across
the network domain boundaries. Note that due to the exploratory nature of the service
requirement list, it is very important to provide a ubiquitous monitoring and measurement
infrastructure to assist in diagnosis, and performance optimization.
An analysis of the requirements reveals a number of significant advantages to the NOBEL
approach. Main features transpiring from this work are cost saving and automation over a
multi-layer integrated data, control, and management plane infrastructure, with more and
more flexibility and inherent capabilities, such as Traffic Engineering and service discovery,
that allow to meet the challenge posed by emerging applications, and to keep a leadership
position in world-class broadband service delivery.


 2.6     Functional ModelingModelling of the Transport and Service
         Platforms
The functional architecture of connection-orientated (CO) layer networks can be described
using G.805, the "Generic Functional Architecture of Transport Networks". The functional
architecture of connectionless-orientated (CL) layer networks can be described using
G.809, the "Functional Architecture of Connectionless Layer Networks".



                                       Page 29 of 168
                             IST IP NOBEL "Next generation
                                                                               Title of the document
                             Optical network for Broadband
                                  European Leadership"                    e1686d81-2994-497c-93f6-
                                                                   b2e913b07f59.docD06_DRAFT_V1
                                                                               6JW2 Andreas zweiter
                                                                        Teil.docD06_DRAFT_V15ejw




A detailed description of the Functional Modeling of the Transport and Service Platforms is
outside the scope of this document: just some guidelines anand the related references are
provided.

                      2.6.1.1 ModelingModelling of Connection- oriented layer
                             networks
Connections are transport entities in CO layer networks that consist of an associated pair
of unidirectional connections capable of simultaneously transferring information in opposite
directions between their respective inputs and outputs. A network connection is a transport
entity in a CO layer network formed by a series of contiguous link connections and/or
subnetwork connections between termination connection points (TCPs).
A subnetwork is a topological component in a CO layer network used to effect routing of a
specific characteristic information, and contains a set of points associated with a
management function within a single CO network layer. A subnetwork connection transfers
information across a subnetwork, and is formed by the association of ports (output of a trail
termination source/input of a trail termination sink) on the boundary of the subnetwork.
Link connections interconnect topologically adjacent sub-networks that have a common
subset of points. The point at which the input of a link connection is bound to the output of
another link connection is a connection point (CP). A set of CPs that areis connected via
link connections to a corresponding set of CPs in a topologically adjacent subnetwork form
a transit group (TG) and the complete set of link connections form a link.

                      2.6.1.2 ModelingModelling of Connectionless layer
                             networks
CL layer networks support any-to-any (a2a) topology constructs. Flows are an aggregation
of one or more traffic units in CL layer networks with an element of common routing. Flows
can be unidirectional or bi-directional, with bi-directional flows consisting of two contra-
directional unidirectional flows. A network flow is a transport entity in a CL layer network
formed by a series of contiguous flows between termination flow points (TFPs).
A flow domain is a topological component in a CL layer network used to effect routing of
specific characteristic information. A flow domain flow is a transport entity that transfers
information across a flow domain, and is formed by the association of ports on the
boundary of the flow domain. A flow domain contains a set of points associated with a
management function within a single CL network layer.
A flow point (FP) is a reference point that represents a point of transfer for traffic units
between topological components. A flow point pool (FPP) is a group of co-located flow
points that have a common routing and an access group is a group of co-located flow
termination functions that are attached to the same flow domain or flow point pool link.
Flow point pool links are topological components that describe a fixed relationship between
a flow domain or access group and another flow domain or access group. Link flows
enable the transfer of information between ports across a flow point pool link.




                                      Page 30 of 168
                             IST IP NOBEL "Next generation
                                                                               Title of the document
                             Optical network for Broadband
                                  European Leadership"                    e1686d81-2994-497c-93f6-
                                                                   b2e913b07f59.docD06_DRAFT_V1
                                                                               6JW2 Andreas zweiter
                                                                        Teil.docD06_DRAFT_V15ejw




 2.7    Methodology for defining migration paths
The emerging of the ASON/GMPLS technology and its potential introduction in the
tTransport Nnetwork (metro/core) creates the problem of how to evolve from the current
legacy network to the new one.
Unless the Nnetwork Pprovider is building a completely new network, this problem
depends on the strategic decisions about the type of network equipments and solutions are
to be introduced and in what evolutionary paths thus satisfying the network evolution
requirements.
The above decisions, such as the time order in which different types of equipment are
introduced and the type of interworking used, will affect subsequent stages of evolution and
may pose networking or interworking constraints.
Appendix III of ITU-T Recommendation G.872 provides information on how a transport
network could evolve to one based on the Optical Transport Network. For example, the two
most important questions are identifying the types of client layer networks to be transported
by the evolving transport network and the type of interworking.
The same basic approach has been adopted by NOBELNOBEL has adopted the same
basic approachobel in defining how a transport network could evolve from the current one
to an innovative one accounting for the emerging requirements.




                                      Page 31 of 168
                                   IST IP NOBEL "Next generation
                                                                                   Title of the document
                                   Optical network for Broadband
                                        European Leadership"                e1686d81-2994-497c-93f6-
                                                                     b2e913b07f59.docD06_DRAFT_V1
                                                                                 6JW2 Andreas zweiter
                                                                          Teil.docD06_DRAFT_V15ejw




 3       Major drivers for the evolution of existing core and metro
         networks
In the last decade nNetwork pProviders largely deployed Sonet/SDH technology in core
and metro transport networks. Furthermore, investments have been made in the last few
years for introducing WDM point-to-point systems in the core long-haul networks.
One of the major overall problems for the evolution of current transport networks seems to
be that there are not enough demands (killer applications) to use/deploy more capacity and
vice-versa that costs are currently too high to fuel a large adoption of new broadband
applications/services (as new applications/services are expected to replace legacy ones
improving the quality perceived by the Customers).
In this context, the evolution of the transport network is likely to be lead by a few
elementary drivers, such as: new network solutions have to optimise the use of resources
(reducing CAPEX); have to reduce the operating costs (reducing OPEX); have to improve
quality, efficiency in providing current and new services (increasing and generating new
revenues).
Furthermore is widely recognized that the traffic in next generation transport networks will
be progressively dominated by data. This is due to the progressive migration of many
applications and services over the Internet Protocol (IP). Given that the statistical
characteristics of data traffic are rather different from those of traditional traffic, for which
TDM networks have been designed, this will pose important technical requirements.
The architectural studies of NOBEL (WP1) are aimed at defining high-level requirements
(technical and business oriented), technologies and solutions for the evolution of core and
metro networks. The access segment (already covered by the IST IP Muse),) will be
considered only from the perspective of the aggregated traffic offered to the metro core.
Nevertheless a cooperationcooperation with IST IP Muse is already established in order to
define an overall architecture for providing end-to-end broadband services.



                                 Core POP


                            Core (Long Haul)




                          Metro POP


                           Core-Metro




                    Hub Nodes


                            Feeder




                                Access


     Figure 333: Examples of network services for storage and residential
                                   Page 32 of 168
     applications
                                         IST IP NOBEL "Next generation
                                                                                                     Title of the document
                                         Optical network for Broadband
                                              European Leadership"                         e1686d81-2994-497c-93f6-
                                                                                    b2e913b07f59.docD06_DRAFT_V1
                                                                                                6JW2 Andreas zweiter
                                                                                         Teil.docD06_DRAFT_V15ejw




Specifically, from a technology viewpoint, NOBEL is mainly focussing on ASON/GMPLS
architectures as they seem to be the most widely recognised solution to match the
emerging requirements, whilst allowing different levels of integration of network
functionality (e.g. L3 routing, and L1-L2 switching) and more flexible control / management
strategies.
Basically the NOBELobel network is providing three main classes of network services (e.g.
VPN L1, L2 and L3). Different services can be provided by separate networks, or different
services may share same network resources.
Figure 48 shows two examples of these service networks; in particular, the upper scenario
represents a generic storage service. The CPE is represented by the data-machine
interface, the internal Storage Area Network collects the traffic coming from the CPEs and
it is the Ffeeder (metro hub) depicted in Figure 4Figure 4Figure 4; in this scenario a layer 1
network service is imagined, so there are optical equipments (OADMS) between the metro
and the core network.
The scenario depicted in the lower section of Figure 4Figure 4Figure 4 represents a typical
residential Internet application where the Ppersonal Ccomputer / Mmodem is the CPE.
Inside the ISP (Internet Service Provider) building it is possible to find the collector of
residential traffic (fFeeder) and, from the downstream side, the point of injection toward of
the collected traffic towards the metro network (PoP metro).

                             Nobel

          Server             gateway/                                                 gateway/
                             extender                                                 extender


                                      metro                  core                 metro
                                                                           ODXC                             Storage
                                              ODXC                                               Storage
                   Storage                                                                                 Subsystem
      Storage       Area                                                                          Area
     Subsystem     Network                                                                       Network
                             metro PoP




 Customer’s Data Center                                                                     SSPr’s Data Center
                                               core PoP
                              Feeder




    CPE
                                ISP




                 access                  metro                   core



                             Nobel
Figure 444: Examples of network services for storage and residential applications




                                                          Page 33 of 168
                             IST IP NOBEL "Next generation
                                                                                         Title of the document
                             Optical network for Broadband
                                  European Leadership"                               e1686d81-2994-497c-93f6-
                                                                              b2e913b07f59.docD06_DRAFT_V1
                                                                                          6JW2 Andreas zweiter
                                                                                   Teil.docD06_DRAFT_V15ejw




An essential starting point of the discussion of the NOBEL vision from a general functional
perspective, is the ITU Recommendation Y.2011, which introduces two strata to describe
the network: Services and Transport Stratum From a functional perspective, the
NobelNOBEL vision is in line with the ITU Recommendation Y.2011 which introduces two
strata to describe the network: Services and Transport Stratum (see Figure 5Figure
5Figure 5).


                                                           Management Plane
                                                            Control Plane
                                                           User Plane


                                    NGN Service Layer


                                                           Management Plane
                                                            Control Plane
                                                           User Plane


                                   NGN Transport Layer


             Figure 555: Functional model of a network [ITU- Rec Y2011]


The transport stratum transports digital information, it interconnects users and/or service
nodes. The transport stratum can be further layered based on the layering principles as
defined in ITU G.805/G.809. For example the transport network can be realized using ETH,
SDH and OTN layers.
The Services stratum provides the users with services for the applications using the
transport layer below for building such services (e.g. for connectivity).
Specifically NOBELobel intends generalised the transport stratum to a resources stratum
(including storage, computing, sensors and other H/W, S/W distributed resources).




                                           APPLICATIONS




                                          SERVICES LAYER




                       MANAGEMENT PLANE                        CONTROL PLANE
                        RESOURCE PLANE (TRANSPORT, COMPUTING, STORAGE, etc)

                              MULTI-SERVICE NETWORK INFRASTRUCTURE




                Figure 666: Functional model of the NOBEL network
                                      Page 34 of 168
IST IP NOBEL "Next generation
                                           Title of the document
Optical network for Broadband
     European Leadership"              e1686d81-2994-497c-93f6-
                                b2e913b07f59.docD06_DRAFT_V1
                                            6JW2 Andreas zweiter
                                     Teil.docD06_DRAFT_V15ejw




        Page 35 of 168
                               IST IP NOBEL "Next generation
                                                                                  Title of the document
                               Optical network for Broadband
                                    European Leadership"                    e1686d81-2994-497c-93f6-
                                                                     b2e913b07f59.docD06_DRAFT_V1
                                                                                 6JW2 Andreas zweiter
                                                                          Teil.docD06_DRAFT_V15ejw




 4        Emerging applications
                                                                                                      Formatted: Bullets and Numbering
 4.1      Introduction
In this section particular attention is paid to emerging applications that are fuelling the
growing of data-traffic. After putting the topic in context, it was we identifyedy that the
capabilities needed to support more and more distributed applications and it is necessary
to start with their characterization. In general, the study take care of we look into the notion
of compounded applications, and in particular of thewe investigatione of exemplary cases
such as the storage, multimedia, and Grid frameworks.
Going from the single application cases towards to most complex applications, i.e. from
fundamental building blocks towards larger systems, the studywe proceed by investigating
the relationship between the complexity of applications and the implications on the required
services as well as on the relevant network infrastructure. It isWe concluded by describing
possible future developments and a short summary.
The research was madeleaded from a more technological point of view; in fact, for each
user application, the effort was concentrated to valuate some parameters were grouped
into the following about the main classes for characterizing the applications.
        Resilience (information about the survivability of an existing connection);
        Provisioning (information about the rate of set-up or tear down a connection);
        Dynamics (information about if/which the parameters of an existing connection that
         can be modified without tear-down, for instance connectivity, QoS parameters
         and/or bandwidth);
        Bandwidth (information about the amount of peak and average bandwidth and its
         granularity);
        QoS (information about whole and latency,differential latency variation, time and
         data integrity on an existing connection);
        Connectivity (connection topology);
        Transparency (tThe primary framing         protocols that can be carried on the
         connection);
        Interface (information about physical interface requirement).
                                                                                                      Formatted: Bullets and Numbering
 4.14.2 Storage applications

         4.1.14.2.1 General Issues
Because of the increasing presence of services in our lives, there is an explosive increase
in the total amount of storage capacity and storage applications. Indeed, these services
permanently extract knowledge from the data stored inside federated ‘knowledge bases’ or
repositories, and even data farms, which are more and more distributed around the world,



                                        Page 36 of 168
                              IST IP NOBEL "Next generation
                                                                                 Title of the document
                              Optical network for Broadband
                                   European Leadership"                      e1686d81-2994-497c-93f6-
                                                                      b2e913b07f59.docD06_DRAFT_V1
                                                                                  6JW2 Andreas zweiter
                                                                           Teil.docD06_DRAFT_V15ejw




and migrate from LAN to extended storage networks using WAN connectivity. In response,
more and more alternativeses to traditional direct attached storage (DAS) technology are
being deployed to make storage devices independent of servers and allow the
consolidation of data accessed from multiple servers.
Before characterizing storage applications it may be interesting (and useful) to define which
particular applications that are the most important from the market point of view. This is not
a market report, but this brief introduction is useful to determine the which storage
applications to consider, independently from the number of storage devices and storage
technologies, see Figure 7Figure 7Figure 7.


                           Source of Revenue by Service Type
                               2001                        2002


                           20%                        19%       29%
                                      41%
                        12%                          20%
                          13%                                   19%
                                 14%                    13%

                         Server Backup               Mirroring, Second Copy
                         Enabling Tools & Services
                         Primary Disk SoD            Others


         Figure 777: Gartner - Storage Services and SSPs: The Insider's View
                              (14 February 2002; USA Market).
From a network point of view, the main network storage applications include:
   1. Remote Bback-up/Rrestore/valuting: Backup to tape/disk located in a remote site
      with infrequent update and latency/distance insensitivity (update interval more than
      one day)
   2. Disaster Rrecovery, with synchronous / assynchronous mirroring where local and
      remote data are always up to date (real time / update interval less than one day)
   3. Business Ccontinuity, with redundant storage and server concept. In case of
      failures on the local site the redundant data centre takes over all functions,
      providing ‘business as usual’ even in case of failures. Example: Symmetrix Remote
      Data Facility (SRDF) providing data availability in minutes for simple and fast
      recovery.
   4. Inter-SAN data migration
   5. Storage on Demand (SoD)
From the network point of view, the Rrestore application is the same application as Bback-
uUp but operates in the complementary direction. The main difference is the lease time,
usually longer for Rrestore than for Bback-up ever since Bback-up may be done by
incremental techniques that don’t transfer the whole data-set, but only the part changed
from the last back-up; the Rrestore must transfer the entire data-set. About the



                                         Page 37 of 168
                             IST IP NOBEL "Next generation
                                                                              Title of the document
                             Optical network for Broadband
                                  European Leadership"                   e1686d81-2994-497c-93f6-
                                                                  b2e913b07f59.docD06_DRAFT_V1
                                                                              6JW2 Andreas zweiter
                                                                       Teil.docD06_DRAFT_V15ejw




requirements (QoS, Bbit-rate ranges, rResilience, pProvisioning, sSecurity and
mManagement) bBack-up and rRestore are very similar and so we decided to combine
themjoin.
It must be noted that synchronous dDisaster rRecovery is latency and distance sensitive
while assynchronous mirroring is not sensitive to these factors. On the other hand, other
familiar applications in the storage networkingstorage-networking field like bBusiness
cContinuity are very similar to the rRemote bBack-up, from the network point of view.
The I-SAN data migration is comprehensive of all the application that moves information
between data-centres. In our opinion, it is not correct to define I-SAN data migration as an
independent application but as a mixture of bBack-uUp, mMirroring (synchronous and
asynchronous) and SoD, so we decided to not consider this macro-application stand-alone.
Other familiar applications in the storage networkingstorage-networking field like bBusiness
cContinuity. (Business Continuity means ensuringEnsuring the continuity or uninterrupted
provision of operations and services.) are very similar to the synchronous mirroring from a
network point of view.
Similar situation for the commercial applications (GDPS and SRDF): for a network point of
view they might be considered as synchronous (or asynchronous) mirroring.
After these considerations, the storage applications, considered in the following from a
network point of view are:
       1.    Remote Back-up / valuting
       2.    Storage on Demand (SoD)
       3.    Asynchronous Mirroring
       4.    Synchronous Mirroring
These four storage application categories will be used in the following requirements
analysis.
                                                                                                  Formatted: Bullets and Numbering
       4.1.24.2.2 Parameters

                     4.1.2.14.2.2.1 Resilience
It is not easy to define generic resilience requirements for storage applications, because
the requested resilience may change with the data at stake. Our assumpsupposition,
confirmed by many recent studies on storage networking, is that the data protected by back
up are less important than the onesthose protected by mirroring, and so we assigned to
mirroring-based applications severer requirements to mirroring applications.
Another situation is the field of Storage on Demand, because the transferred data may be
the only existing copy and in this casecase,, a high level of availability for these
applications is needed, even if usually the outsourced data by SoD are the least important
for a company. After these considerations the values of some parameters are summarized
in Table 7Table 7Table 7.




                                      Page 38 of 168
                              IST IP NOBEL "Next generation
                                                                                                                                                                                                                                                                            Title of the document
                              Optical network for Broadband
                                   European Leadership"                                                                                                                                                                                                                 e1686d81-2994-497c-93f6-
                                                                                                                                                                                                                                                                 b2e913b07f59.docD06_DRAFT_V1
                                                                                                                                                                                                                                                                             6JW2 Andreas zweiter
                                                                                                                                                                                                                                                                      Teil.docD06_DRAFT_V15ejw




                                                                                                                Resilience




                                                                                                                                                                       Block prob of restoration
                                                                                                                             Max recovery time [ms]
                                             Application




                                                                                                                                                                                                                    Max outage time
                            Back-Up / Vaulting                    4-9               Min availability                                   0?                                                           1 day
                            Storage on demand                     5-9                                                       100 ?                                                                   hours
                            Asynchronous Mirroring                5-9                                                       100 ?                                                                   hours
                            Synchronous Mirroring                 5-9                                                                  0?                                                           hours


                Table 777: Resilience parameters to storage applications
The column “Max recovery time” has a particular importance in particular for the
applications usually carried by Fibre Channel (FC). In particular FC equipments are able to
notice micro-interruptions of the signal, less than the common recovery time of 50 ms for
(SDH), and prime cause a chain of events, if they are carried over a SDH infrastructure,
that carry leads to an outage of the service for up to tens of seconds, so every interruption,
even if for very short time, are to be considered as an unavailability event.
                                                                                                                                                                                                                                                                                                Formatted: Bullets and Numbering
                       4.1.2.24.2.2.2 Provisioning
For the provisioning, it is important to split the considered storage applications into two
main categories:
      The scheduled applications (i.e. back-up/restore/vaulting);
      The non-scheduled applications (i.e. mirroring and SoD).
                                                                                                                                                                                                                                                                                                Formatted: Bullets and Numbering


                                                                                                       Provisioning
                                                                                                                                                                                                                                      Max blocking probability
                                    Application




                                                                                                       Max tear-down time
                                                                  Max set-up time




                                                                                                                                                                                                   Max lease time
                                                                                                                                                      Min lease time




                          Back-Up / Vaulting                min                     min                                     min                                        hours ?
                          Storage on demand                 sec                     sec                                                                     0 perm. ?
                          Asynchronous Mirroring            sec                     sec                                                                     0 perm. ?
                          Synchronous Mirroring             min                     min                                     hours perm. ?


          Table 888: Provisioning time requirements for storage applications
                                                                                                                                                                                                                                                                                                Formatted: Bullets and Numbering
                       4.1.2.34.2.2.3 Dynamics
About For bBack-uUp applications, it is not so important to change the connection
parameters during the link lease time. The same consideration is valid about for
sSynchronous mirroring; in particularparticular, for the this application it is important to note



                                                     Page 39 of 168
                              IST IP NOBEL "Next generation
                                                                                                                                       Title of the document
                              Optical network for Broadband
                                   European Leadership"                                                                            e1686d81-2994-497c-93f6-
                                                                                                                            b2e913b07f59.docD06_DRAFT_V1
                                                                                                                                        6JW2 Andreas zweiter
                                                                                                                                 Teil.docD06_DRAFT_V15ejw




how is the traffic is generated by it. It is a sort “comb” of traffic spikesspikes; each of them
has a very brief length of time (a few ms) and a very high bandwidth (hundreds of Mbit/s). It
is impossible to schedule these spikes, and there is not enough time to set-upset up a
connection when a spike of traffic occurs.; so,Therefore, the best way to operate is to have
a constant connection and, where possible, shared it between some different customers,
while minimizingminimising the blocking probability.
About For asynchronous movement of data (i.e. aAsynchronous mirroring and SoD), it may
be interesting to shape the traffic and to modify the QoS levels in operation of the
importance of data.

                                                                  Dynamics
                                        Application




                                                          Bitrate management




                                                                                                        Connectivity mgmt
                                                                                   QoS management




                             Back-Up / Vaulting       N                        N                    N
                             Storage on demand        Y                        Y                    N
                             Asynchronous Mirroring   Y                        Y                    N
                             Synchronous Mirroring    N                        N                    N


              Table 999: Dynamics requirements for storage applications


                                                                                                                                                           Formatted: Bullets and Numbering
                      4.1.2.44.2.2.4 Bit-Rate
Synchronous mirroring is surely the storage application that needs the most important
bandwidth share. In particular, it is impossible to have a synchronous (mirroring)
application using at a level of bandwidth below 100-200 Mbit/s.
For the reason discussed about dynamics parameters, it is impossible to predict how the
spikes of traffic occurs and set up a dedicated connection specifically for each spike.
Back-u Up usually is usually a scheduled activity, and when it is set-upset up, it needs of a
fixed connection. For this reasons it is nonsense irrelevant to define max
incremental/decremental time for these applications. The situation is different considering
for asynchronous mirroring and storage on demand. For these applications, the bandwidth
may change in operation of the quantity of data during the lease time of the connection.




                                       Page 40 of 168
                              IST IP NOBEL "Next generation
                                                                                                                                                                                                                                                                 Title of the document
                              Optical network for Broadband
                                   European Leadership"                                                                                                                                                                                                e1686d81-2994-497c-93f6-
                                                                                                                                                                                                                                                b2e913b07f59.docD06_DRAFT_V1
                                                                                                                                                                                                                                                            6JW2 Andreas zweiter
                                                                                                                                                                                                                                                     Teil.docD06_DRAFT_V15ejw




                                                                                                            Bitrate [Mb/s]




                                                                                                                                                                                                   Max Decrement time
                                   Application




                                                                   Max Nominal Bitrate




                                                                                                                                                         Max Increment time
                                                                                                            Min Dynamic range




                                                                                                                                                                                                                                                BR Granularity
                        Back-Up / Vaulting                        400 const N.A.                                                                                              N.A.
                        Storage on demand                        1000                                                           sec                                           sec
                        Asynchronous Mirroring                    400                                                           sec                                           sec
                        Synchronous Mirroring                    2000 const N.A.                                                                                              N.A.


              Table 101010: Bitrate requirements for storage applications
                                                                                                                                                                                                                                                                                     Formatted: Bullets and Numbering
                      4.1.2.54.2.2.5 Quality of Service (QoS)
For this item, it is important to note that some parameters are carried solely by L1
connection, and for these ones we defined only the Bit Error Rate (BER). For the
applications that generate traffic carried also on L3, we have to define a packet loss.
The most important consideration appearing in this section of the table is the max latency
requirement forin particular referred to the synchronous mirroring, Table 11Table 11Table
11. It is very important that the latency remains below a few ms otherwise the storage
application collapses.
SEver since every storage application has mechanisms to reorder frames, it is not so
important the difference of latency between frames is not so important, being understood
that it remains below a certain limit; otherwise, the receipt receiver will need of a very big
buffer to store the frames waiting to be reordered.

                                                                                                                                    QoS
                                                                                                                                                                                                                        Packet Loss (Layer 3)
                                                                                                                                Max diff. Latency [ms]
                                                 Application




                                                                                                                                                                              Max BER (Layer 1+)
                                                                                         Max latency [ms]




                           Back-Up / Vaulting                      N.A.                                     N.A.                                                                                                  1%
                           Storage on demand                                             10                                               1 1E-12                                                                 1%
                           Asynchronous Mirroring                                  100                                          10 N.A.                                                                           1%
                           Synchronous Mirroring                                                3                                         1 1E-12


           Table 111111: Some QoS requirements for storage applications
                                                                                                                                                                                                                                                                                     Formatted: Bullets and Numbering
                      4.1.2.64.2.2.6 Connectivity – Transparency – Interface
These parameters (connectivity, transparency and interface) are quite implicit in previous
paragraphs considerations. In particular, storage services are point- – to- – point.
ItHowever, it is possible to have some services that have onea source and multiple
recipients, such as a back up done over multiple sites.



                                                          Page 41 of 168
                                 IST IP NOBEL "Next generation
                                                                                             Title of the document
                                 Optical network for Broadband
                                      European Leadership"                               e1686d81-2994-497c-93f6-
                                                                                  b2e913b07f59.docD06_DRAFT_V1
                                                                                              6JW2 Andreas zweiter
                                                                                       Teil.docD06_DRAFT_V15ejw




From the point of view of the network that serves the storage applications, the service
described is seen like two separate applications asking for two different network services.
About With regard to transparency, synchronous mirroring is the only applications that
necessarily needs an L1 connection ever since it has strictly low-latency requirements.
Back- up, and, in particular, synchronous mirroring is usually carried over Fibre Channel;
for the asynchronous applications, IP/GbE routing/transport protocols are allowed.
It is important to note that in the majority of storage applications (in particular when Fibre
Channel is adopted) layer 2 functionalities are provided at the customer site only, so no
layer 2 functionalities in the network are necessary. The layer required for carry storage
application may be essentially L3 (for applications that don’t need severe performance
requirements and L1 for applications that need low latency, unavailability and jitter.



                                               Connectivity    Transparency      Interface
                             Application




                   Back-Up / Vaulting         p2p             L1-L2-L3        FC/IP/GbE
                   Storage on demand          p2p             L1-L2-L3        IP/GbE
                   Asynchronous Mirroring     p2p             L3              IP/GbE
                   Synchronous Mirroring      p2p             L1              FC/WDM


  Table 121212: Connectivity, transparency and interface requirements for storage
                                    applications
                                                                                                                 Formatted: Bullets and Numbering
                       4.1.2.74.2.2.7 Security
In general, storage applications generate a type of data-traffic particularly valuable and
therefore the level of security must be high. In any case, it is possible to split the issue into
two important fields of security applications (edge and channel).
The security techniques adopted strictly depend on the technology used for carry traffic
(transport column).
      The edge security that includes AAA (Authentication, Authorization and
       Accounting), IDS (Intrusion Detection Systems), firewalling;
      The channel security that includes traffic/data segregation and cryptography.


             Storage Application             Transport         Edge              Channel

                                                              Firewall        VPN IP, VLAN
             Back-uUp/vVaulting                     IP
                                                                AAA              IPsec




                                            Page 42 of 168
                              IST IP NOBEL "Next generation
                                                                                       Title of the document
                              Optical network for Broadband
                                   European Leadership"                      e1686d81-2994-497c-93f6-
                                                                      b2e913b07f59.docD06_DRAFT_V1
                                                                                  6JW2 Andreas zweiter
                                                                           Teil.docD06_DRAFT_V15ejw




                                                      Firewall     VPN IP, VLAN
                                             IP
                                                        AAA           IPsec
            Storage on Demand
                                            WDM                  Traffic segregation
                                                         -
                                            SDH                   (Cryptography)
                                                      Firewall     VPN IP, VLAN
            Asynchronous Mmirroring          IP
                                                        AAA           IPsec

                                            WDM                  Traffic segregation
            Synchronous mMirroring                       -
                                            SDH                   (Cryptography)

Table 131313: Authentications, authorizationauthorisation, and accounting
requirements
At this point may be useful to map the security requirements described above into some
classes of security.
                                                                                                           Formatted: Bullets and Numbering
       4.1.34.2.3 Conclusions on Storage Applications
Analysing the parameters singled out in our study, it is obvious to note that SoD may be
split into two services: The first one carried on Layer 1 and the second one carried on
Layer 3. While characterizingcharacterising four network storage applications may be
complex, one idea may be to define three application categories:
       1.      Back-up/Vaulting;
       2.      Asynchronous applications (includes aAsynchronous mMirroring and the
               section of SoD carried in asynchronous mode on L3)
       3.      Synchronous applications (includes sSynchronous mirroring and the section
               of SoD carried in asynchronous mode on L1)

At this point it is possible to consider the four storage applications together, as described in
this document, or to merge this four applications into the three categories described above,
losing however some details.
                                                                                                           Formatted: Bullets and Numbering
       4.1.44.2.4 Important proprietary storage applications: IBM
             GDPS and EMC SRDF
IBM GDPS (“Geographically Dispersed Parallel Sysplex”)
GDPS is a multi-site application and data availability solution designed to provide the
capability to manage the remote copy configuration and storage subsystem(s), automate
Parallel Sysplex (proprietary implementation of disaster/recovery – business continiuity of
IBM) operational tasks, and perform failure recovery from a single point of control, thereby
helping to improvinge application availability. With GDPS many different processors may
be interconnected through a coupling connection facility, which allows them to
communicate with each other and with data stored locally.
Since all processors in a Parallel Sysplex must operate synchronously with each other,
they all require at least an 8 Mbit/s multimode fibre link or better a DWDM connectivity
[Skanova].



                                        Page 43 of 168
                                IST IP NOBEL "Next generation
                                                                                            Title of the document
                                Optical network for Broadband
                                     European Leadership"                           e1686d81-2994-497c-93f6-
                                                                             b2e913b07f59.docD06_DRAFT_V1
                                                                                         6JW2 Andreas zweiter
                                                                                  Teil.docD06_DRAFT_V15ejw




EMC2 SRDF (“Symmetrix Remote Data Facility”)
SRDF generates a mirror image of the data at logical volume level in one or more remote
Symmetrix systems. For When using SRDF in a wide area distances range from 100km to
trans-oceanic or/ continental lengths. Typically, ATM, T3/E3, T1/E1 or IP connectivity are is
offered by carriers for this application [EBU I35].
                                                                                                                Formatted: Bullets and Numbering
 4.24.3 Multimedia applications
Multimedia technologies provide effective means for business, social/educational, and
entertainment applications across time and space. In essence, multimedia is a fusion of
multiple types of data sources used to acquire, process, transmit, store, and utilize
information. In this context, a special attention is brought in this document on some
exemplary applications that include:
      Video-streaming
      Voice over IP
      Video conference
      Video data applications
      On-line gaming
      Tele-medicine application
                                                                                                                Formatted: Bullets and Numbering
       4.2.14.3.1 Background
Multimedia applications are highly related to compression technology. Tables with the
essential content of Table 1 can be found in several documents. The material refers back
to early Mpeg-2 documents around 1994(!), but is recurring with updated comments up to
2002. The stability of the data illustrates that the basic performance of a coding technology
lies basically in its definition and algorithms, and less in its implementation. The latter more
affects the size and cost of the equipment needed to obtain the performance.
       The Test subgroup has defined a few example "Sweet spot" sampling dimensions
       and bit rates for MpegPEG-2: These numbers may be too ambitious. Bit rates of 3,
       6, and 8 Mbit/sec respectively provide transparent quality for the above application
       examples when generated by a reasonably sophisticated encode (1996).


    Size     Bitrate [Mb/s]                                   Application
  576x352         2-3         Equivalent to VHS quality. Intended for film source video. Half horizontal
   (24Hz)                     601(HHR). Looks almost broadcast NTSC quality.
  576x544         4-6         PAL broadcast quality 544 samples matches the width of a 4:3 pic ture
   (30Hz)                     windowed within 720 sample/line 16:9 aspect ratio via pan&scan
  576x704         6-8         Full CCIR 601 sampling dimensions, “Studio quality” (DV, DVD)
   (30Hz)

 Table 141414: Typical coded bandwidths for MPEG-2 for different quality classes


                                           Page 44 of 168
                             IST IP NOBEL "Next generation
                                                                               Title of the document
                             Optical network for Broadband
                                  European Leadership"                    e1686d81-2994-497c-93f6-
                                                                   b2e913b07f59.docD06_DRAFT_V1
                                                                               6JW2 Andreas zweiter
                                                                        Teil.docD06_DRAFT_V15ejw




The codec (COmpression and DECompression) compress the data to reduce the size of
the information transfer. There is a flurry of algorithms but the most common used for
digital video and audio are probably the Mpeg algorithms, which are defined in a set of ISO
standards:.
   The Mpeg-1 was designed for playback from a CD-Rom and Mpeg-1 Layer 3 is widely                Formatted: Bullets and Numbering
    used for audio distribution over the Internet (mp3).
   Mpeg-2 supports higher bit-rates and is today used by DVD players and for digital TV
    broadcasting (DVB).
   Mpeg–4 is designed for error prone transmission channels and to give better quality at
    lower bit-rates. H.264 is codec standardised by ITU, which is Mpeg-4 compliant.
Investigations [EBU343] show that that H.264 compression will produce higher quality
efficiency than Mpeg2, at all quality levels. For PAL broadcast quality standard definition
TV (SDTV) the gain can be more then 50%. DivX5 is one specific implementation of Mpeg-
4.

NOBEL should estimate the dominant technology/standard within the main NOBEL
timeframe and stick to that technology assumption. The suitable choice for the medium
NOBEL timeframe of 3-5 years is clearly Mpeg-4 AVC (Mpeg-4 track 10), also known as
H.264. This is an emerging technology (leading Mpeg-4 family) today, and can be assumed
to have a strong market penetration in this timeframe. It is not sensible to further assume
that yet another generation of technology will be dominant at the same time in the same
market. Thus, the average bandwidth gain for NOBEL compared to Mpeg-2 (today’s
mature technology) can be estimated to a factor of 2 from most papers on H.264 [EBU I34].


In addition to considering the dominant technology generation, the actual end-user
equipment to be served by the transport network is likely to consist of an average over the
last 3-5 years of technology (of that generation), having different creation times along the
learning curve. As the saturation point of a technology generation approaches, the next
generation becomes important, and the market share for the previous generation probably
has already passed its maximum. Then, the average equipment age is probably around 2
years, and the next generation is already emerging. Thus, the saturation point will never be
the average performance point, and the saturation point of the learning curve cannot be
used as a basis for transport network dimensioning. The average technology level (in a
given generation) should rather be taken as the average equipment age of that generation,
probably ca 2 years.

It is not easy to find a bandwidth equivalent to this number. However, lLooking at Mpeg-4, it
is often stated, e.g. in [EBU I35] that the technology uses 2-4 times the algorithmic
complexity and up to ten times the memory requirements compared to Mpeg-2. This
means that, with Moore’s law of a factor of 2 in processor capacity gain per 18 months,
Mpeg-4 will “use up” about 2-3 years of processor development in order to (at the same
price-point) reach the claimed bandwidth gain of ca 50% over Mpeg-2. Thus, it is not
unreasonable to assume that an average equipment age of 2 years will correspond to a
bandwidth span of a factor of 2, and thus the correction factor for average equipment age
would be ca 1.5. This might be more pronounced early in the development, as the learning




                                      Page 45 of 168
                              IST IP NOBEL "Next generation
                                                                                Title of the document
                              Optical network for Broadband
                                   European Leadership"                   e1686d81-2994-497c-93f6-
                                                                   b2e913b07f59.docD06_DRAFT_V1
                                                                               6JW2 Andreas zweiter
                                                                        Teil.docD06_DRAFT_V15ejw




curve typically is steeper. Thus, the initial correction for Mpeg-4 (emerging technology)
would be higher than what we see today for Mpeg-2 (mature technology).

Therefore, NOBEL should estimate the dominant technology/standard within the main
NOBEL timeframe and stick to that technology assumption. The suitable choice for the
medium NOBEL timeframe of 3-5 years is clearly Mpeg-4 AVC (Mpeg-4 track 10), also
known as H.264. This is an emerging technology (leading Mpeg-4 family) today, and can
be assumed to have a strong market penetration in this timeframe. It is not sensible to
further assume that yet another generation of technology will be dominant at the same time
in the same market. Thus, the average bandwidth gain for NOBEL compared to Mpeg-2
(today’s mature technology) can be estimated to a factor of 2 from most papers on H.264
[EBU I34].
The error resilience of a codec can be improved by using tools for data recovery and error
concealment. This is included in the Mpeg-4 standard. Data recovery is used to detect and
correct bit-errors by using a coding overhead. The technique is important when the channel
is characterised by random errors. The main purpose of error concealment is to limit the
impact of packet loss on the image quality by using the information of adjacent frames.

The conclusion here is that even early research claims more tend to describe the
saturation point than average market performance at that time and (as expected from
research) are quite forward looking. Thus the numbers quoted already point to the future of
that technology, and have to be applied to real life situations considering several “diluting”
factors, applicable to each generation in itself:

•      Market penetration (of that generation)

•      Average age of equipment on market (of that generation)

•      Spread of implementation level of equipment (consumer to professional)

After a given technology generation has gone into saturation, and the next generation
emerges, giving a “step function” in performance, it will take (1) some time for this
generation to penetrate the market and (2) that specific generation will in turn keep
improving according to its learning curve, see Figure 8..




                                       Page 46 of 168
                              IST IP NOBEL "Next generation
                                                                                   Title of the document
                              Optical network for Broadband
                                   European Leadership"                     e1686d81-2994-497c-93f6-
                                                                     b2e913b07f59.docD06_DRAFT_V1
                                                                                 6JW2 Andreas zweiter
                                                                          Teil.docD06_DRAFT_V15ejw




Figure 888: The evolution of standard video compression systems. A pattern of
         development cycles occurs, which results in long-run continous gains in
         efficiency [EBU I353].
Regarding the traffic type, as well prepared content by the best possible equipment is quite
better coded in bandwidth and quality compared to either automated procedures or real-
time traffic, the traffic mix can be an important influence on the average bandwidth. If
practically all traffic is high-end movies (VoD), the average coding is probably of much
better quality than if there is a substantial part of real time and mass-coded content. An
example of the latter may be if large archives of e.g. historical material are made available,
or if video becomes a popular trade in peer-to-peer networks.
The content of a scene or movie also strongly affects the bandwidth. It is one of the
approaches of modern coding to adapt the coding to the rate of change of the content by
numerous techniques (all of which demand considerable computing complexity).
                                                                                                       Formatted: Bullets and Numbering
       4.2.24.3.2 Video streaming (video-on-demand, VoD, video broadcast,
              and IP-TV)
Video on demand (VoD) is the video streaming to a single user. IP-TV is video streaming to
multiple users (multicast). Standard tools that are used to support the transmission and to
relax on the requirements of the network are data compression, error concealment and
using a memory buffer at the receiving end. A buffer at the receiving end increases the
average latency but it makes the transmission more tolerant to jitter (latency variations). In
applications with bi-directional real-time communication (telephony and video conference)
the buffer size is limited by the requirements of having a low latency. For VoD and IP-TV
the information, transfer is unidirectional and it is in principle possible to use large a buffer
(seconds).
Digital video broadcasting (DVB)
The traditional way of using the video media is to watch TV. The way TV programmes are
distributed to end-users has evolved over time, from wireless terrestrial, to transmission
through coaxial cables, and satellite transmission. Independently of distribution, form there
is in parallel an ongoing changeover from analogue transmission to digital transmission.
The DVB-project and the European Broadcasting Union (EBU) drive development of
procedures and standards for broadcasting digital TV. The DVB-Project is an industry-led
consortium of over 260 broadcasters, manufacturers, network operators, software
developers, regulatory bodies and others in over 35 countries committed to designing




                                        Page 47 of 168
                              IST IP NOBEL "Next generation
                                                                                Title of the document
                              Optical network for Broadband
                                   European Leadership"                   e1686d81-2994-497c-93f6-
                                                                   b2e913b07f59.docD06_DRAFT_V1
                                                                               6JW2 Andreas zweiter
                                                                        Teil.docD06_DRAFT_V15ejw




global standards for the global delivery of digital television. DVB standards are published
by the European Telecommunications Standards (ETSI).
The DVB-project represents an industry working already today with the broadcast of digital
video. The standards are published by ETSI-. Standards for DVB services over IP based
networks are being developed [ETSI1]. In this context, IP-TV is just a new distribution form
for a well-defined and established service. Standards for DVB services over IP based
networks are being developed [ETSI1].
The only allowed codec in DVB so far is mpeg2 and single SDTV channels are
compressed down to about 3-5 Mbit/s [EBU I34]. For terrestrial and satellite distribution the
channels are clustered in groups, so-called channel multiplexes, having gross bit-rate of
225 Mbit/s, corresponding to the capacity of the transmission channel (e.g. a satellite
transponder or a terrestrial carrier frequency). The channels are compressed and
statistically multiplexed onto a channel multiplex with a single device. The channels are
normally coded to have constant quality and the bit-rate of individual channels within a
multiplex varies with time and content.
Flat screens and high-definition TV (HDTV) are trends that raise demands on increased
bandwidth. Development and evolution of codecs will relax the demands on the bandwidth
and work has started to include mpeg4 into the DVB standard. However, changes in the
bit-rates of TV channels will only effect how a channel multiplex is used. Today up to six
SDTV channels (PAL broadcast quality) are combined on one multiplex. In the future, a
multiplex could carry a mixture of SDTV and HDTV channels. It is also possible to use the
bandwidth of a multiplex to transmit a fewer channels that are more resilient to errors.
A transport service that could be interesting to digital TV broadcasters is the transmission
of a channel multiplex.


Internet-TV
An application that has emerged with broadband access and DSL is streaming of video
directly over the Internet. There are so far no standards for doing this, the market is
dominated by two propitiate streaming platforms: Real10 and WindowsMedia9. A big
difference compared to the DVB work is that the transmission channel is unspecified. The
focused is therefore on giving high quality at lower bit-rates and to find solutions that are
resilient to transmission errors such as jitter and packet loss.
State-of-the-art techniques for VoD, that are available today, have been implemented into
Real10 and WindowsMedia9. Both use codecs similar to mpeg-4 (but not mpeg-4
compliant) that have tools for error concealment included. The tolerance towards packet
loss has been dramatically increased by extending the buffer size to the order of 10s. This
allows for retransmission of lost packages. Note that a large buffer does not imply a long
start-up time since playback can start before the buffer is full (e.g. turbo play in RealOne).
Real10 claims IP-TV at 1Mbit/s and HDTV at < 5 Mbit/s with real-time SW-decoding on a
PC. These tools develop continuously and the future requirements on the network
performance will probably be relaxed because of this.
There is a difference in the requirements on packet loss between unicast and multicast.
Multicast is more sensitive to packet loss because it is not possible to use retransmission




                                       Page 48 of 168
                              IST IP NOBEL "Next generation
                                                                                 Title of the document
                              Optical network for Broadband
                                   European Leadership"                    e1686d81-2994-497c-93f6-
                                                                    b2e913b07f59.docD06_DRAFT_V1
                                                                                6JW2 Andreas zweiter
                                                                         Teil.docD06_DRAFT_V15ejw




largely. Requests of retransmission from a large number of destinations would overload the
servers.
For streaming video directly over the Internet a large span of bit-rates are of interest
ranging from slow connections limited by the accessible bandwidth (down to 100 kbit/s) to
HDTV signals.


Bit-Rate
There are two trends in the distribution of digital video. Existing players on the television
industry represent the first trend: e.g. production companies, TV channels, and equipment
manufacturers. These use that DVB standard for broadcast of video over several different
transmission channels, e.g. terrestrial wireless, satellite, CATV or IP networks. Currently
mpeg2 compression to about 3-5 Mbit/s is used for standard definition TV. Increased
demands on the picture quality due of flat screen are expected and HDTV is an important
emerging application. At the same time, codecs develop with time and work is ongoing in
order to include Mpeg4-AVC in the DVB standard.
For streaming video directly over the Internet a large span of bit-rate are of interest ranging
from slow connections limited by the accessible bandwidth (down to 100 kbit/s) to SDTV
signal. The existing tools for streaming video over the Internet (WindowsMedia and Real)
use mpeg4 compression.
To have an idea of the requested bandwidth for an SDTV channel compressed with mpeg4
codec it may be useful these considerations:
2 h of MPEG-4 video (+ audio) take up about 1.5 GB
1.5 GB = 1.5 x 8 = 12 Gbit
The 8b/10b encoding and the protocol overhead add about 30%  12 Gbit + 30% =
15.6 Gbit
                   15.6Gbit
The bandwidth is            = 2.16 Mbit/s
                   7200 sec
Other indications pointing in the same direction are the EBU [EBU I35], claiming mpeg4-
AVC being more then 50% efficient than mpeg2 (results in about 1.5-2.5 Mbit/s per
channel). RealNetworks claim that Real10 is capable of DVD quality video at 80% lower
bit-rate than mpeg-2 (around 1Mbit/s) and HDTV at less then 5 Mbit/s. RealNetworks use
their proprietary codec figures may of course be stretched to put Real10 in a favourable
position.
There is not one single required bit-rate for video streaming; instead there is a large
variation depending on application and on time horizon considered. As brief and compact
summary of how the different factors previously discussed influence the bit-rate, we
propose:
      “Saturation point” to nominal broadcast quality: 1.5
      Average equipment age: 1.5
      Average equipment implementation; 1.5




                                       Page 49 of 168
                             IST IP NOBEL "Next generation
                                                                               Title of the document
                             Optical network for Broadband
                                  European Leadership"                    e1686d81-2994-497c-93f6-
                                                                   b2e913b07f59.docD06_DRAFT_V1
                                                                               6JW2 Andreas zweiter
                                                                        Teil.docD06_DRAFT_V15ejw




      Technology generation: MPEG-4 vs. MPEG-2: 0.5
      Full CIR to HD resolution: 4
      HD to Cinema resolution: 4
      Entertainment/studio/DVD quality for content vs. nominal broadcast quality: 1,5


Applying these to some standard formats digital video for the NOBEL time horizon (5-10
years) assuming MPEG4 gives:
      PAL broadcast: 1.5 -3 Mbit/s
      DVD/entertainment: 2.5 – 5 Mbit/s
      HDTV: 6 – 12 Mbit/s

In parallel there will also be a demand for “best effort” streaming over the Internet using
residential Internet connections without guarantees for any QoS. The bit-rate will be limited
by the connection and from the perspective of the end-user it would be preferable to have
best quality the connection can support. Typical bit-rates for this application are 200-500
kbit/s but almost any bit-rate the Internet (e.g. cheap, flat- rate) connection can support
ranging from 100 kbit/s and upwards could be used.
QoS
Important parameters for describing the quality of an IP connection are the packet loss, the
latency and the variations in latency (jitter). The presence of incumbent applications forces
residential video application to have a high level of QoS. However, both VoD and video
broadcast include only one-way communication, with low the demands on the latency and
this makes it possible to include a large buffer at the receiving end. The maximum latency
varies between different cases, in case of a live transmission, it should be low (e.g. when
watching football you don’t want to hear your neighbour cheer before you see the goal) but
a couple of seconds may be acceptable. In other cases, a latency of 20s is not a problem
and there is little need for extending the buffer further. The advantage of a buffer
compensates for jitter and the buffer time can be used for retransmission – at the
application level – of lost packets.
There is a difference in the requirements on packet loss between unicast and multicast.
Multicast is more sensitive to packet loss because it is not possible to use retransmission
largely. Requests of retransmission from a large number of destinations would overload the
servers.




                                                 Packet      Latency Jitter
                                                  loss
              MPEG2 based applications             10-6        N/A     40 ms




                                      Page 50 of 168
                              IST IP NOBEL "Next generation
                                                                                 Title of the document
                              Optical network for Broadband
                                   European Leadership"                    e1686d81-2994-497c-93f6-
                                                                    b2e913b07f59.docD06_DRAFT_V1
                                                                                6JW2 Andreas zweiter
                                                                         Teil.docD06_DRAFT_V15ejw




               In particular DVB                                (10s)
               MPEG-4 based applications           0.5%         N/A      40 ms
                                                                (10s)
               MPEG-4 based applications            5%          N/A       N/A
               retransmission (e.g. Real 10
                                                                (10s)     (1s)
               and WM9)                                                                              Comment [POA1]: Page: 17
                                                                                                      Add section that defines the common format
                                                                                                     we are using for the numbers; e.g. SDTV,
Table 151515: Important QoS requirements for multimedia applications                                 broadcast quality etc. Possibly supplemented by
                                                                                                     table that cross-refers to other services such as
Without retransmission packet loss must be below 0.5%, otherwise the video quality is                HDTV, video-chat.
dramatically reduced. For the streaming platforms, that has variable bit-rate and tools for
retransmission the packet loss as high as 5% can be acceptable.
Summary of QoS for video applications. For the DVB applications the values for packet
loss and jitter are according to standards. Values for Internet TV refers to typical values for
Real10 and WM9. Last three lines quotes estimated performance of codecs in term of bit-
rate and packet loss. The jitter and latency specification will depend on the application.
Connectivity – Transparency – Interface
Video on demand and video download are typically point-to-point applications. On the
contrary, video broadcast is a multi-user application and so it needs multi/broadcast
connections. Videoconference may be both point-to-point and multicast. For all the
applications (since they are addressed to a residential market) are carried on L3.
Security
The main security issue may be the frauds. In effect, it is possible that someone could
intercept traffic to capture video without paying for it. To prevent this problem, cryptography
and authentication are essential; accounting may also be important for billing.
Video applications will be delivered in many cases from trusted sites, and in such cases, it
is not necessary to filter the data stream for e.g. viruses and Trojan horses. When
streaming from unknown sources this could become a problem that calls for the usage of
e.g. IPsec, packet filtering and IDS (Intrusion Detection Systems).
In our opinion, these considerations are valid for both VoD and IP-TV.
                                                                                                     Formatted: Bullets and Numbering
       4.2.34.3.3 Voice over IP & video chat
In applications where two or more users communicate in real-time the requirement on the
latency becomes much more stringent. If the latency is long, the users will have to wait for
the other parties respond and this makes it difficult to communicate in a natural way. There
are two main protocols for signalling and call control of real-time multimedia
communications over packet-based networks, SIP defend by IETF (RFC3261) and H.323
suite defined by ITU SG 16. These two protocols represent different approaches to solving
the same problem and can deliver close to the same set of services. H.323 is more
comprehensive while SIP is more “light weight”, e.g. in H.323 the CODECcodec has to be
approved and standardised while SIP allows any CODECcodec.
One strong drive for VoIP comes from residential users who want to reduce their phone bill
either by making free calls over Internet or using a commercial VoIP service. In the latter



                                       Page 51 of 168
                             IST IP NOBEL "Next generation
                                                                             Title of the document
                             Optical network for Broadband
                                  European Leadership"                  e1686d81-2994-497c-93f6-
                                                                 b2e913b07f59.docD06_DRAFT_V1
                                                                             6JW2 Andreas zweiter
                                                                      Teil.docD06_DRAFT_V15ejw




case the client use the broadband connection to connect to a gateway where calls enter
the ordinary telephone network. VoIP will bring in competition and thereby lowering the
price. Another important drive comes from enterprise customers wanting to use one single
infrastructure for data and voice traffic.
Resilience
The requirements on the resilience from paying enterprise and residential costumers are
high, similar to that of POTS today. For the best-effort applications the requirements are
lower, in the same order of the Internet connections from ISPs today.
The requirements on the resilience are approximately the same for all residential
applications so VoIP and video-chat are not different from VoD and IP-TV in this respect.
We propose a min availability of 99-99.5% and a maximum outage time of up to 12 hours
(see VoD and IP-TV).
Bit-rate
The requirements are closely related with the codecs that are used. For IP telephony
G.711 (standard 64 kbit/s), G.723 (5.3/6.3 kbit/s) and G.729 (8 kbit/s) are commonly used
but e.g. skype uses their own codec. IP telephony can also offer better experience than
POTS by implementing wideband codecs that encode the analogue signal with higher
bandwidth than 3.1 kHz.
In the case of video conferencing and chat the options are the same as in the case of video
streaming., i.e. in particular mpeg-4 based codecs. A desktop video chat application could
use e.g. 64 kbit/s up to 400 kbit/s and high-quality videoconference service from 400kbit/s
to 1 Mbit/s. On the NOBEL timescale we assume mpeg-4 based codecs. There will be a
demand for a broad range of video qualities and picture resolutions, limited to what the
access connection can support up to broadcast PAL quality. In addition to the factors
considered in the case of video streaming
   Scene simplicity. The video information is in most cases a talking head which is a           Formatted: Bullets and Numbering
    simple scene, easily to compress. Compared to normal PAL broadcast the bit-rate
    requirement is to 0.5.
   Real-time coding penalty. The compression is performed in real-time on a client PC.
    Compared to professional equipment used by TV broadcasters this compression is less
    effective which increases the bandwidth.
For a videoconference having a quality similar to PAL broadcast quality the bit-rate
requirement on the NOBEL time-frametime frame is 1.5-3.0 Mbit/s ,s, assuming 2.0 real-
time coding penaltypenalties.


QoS
Latency requirements are similar for audio and video communication. ETSI TIPHON
[ETSI2] has defined three speech QoS classes and related them to maximum latency,
Table 16Table 16Table 16.


                                                       Max latency




                                     Page 52 of 168
                              IST IP NOBEL "Next generation
                                                                                 Title of the document
                              Optical network for Broadband
                                   European Leadership"                    e1686d81-2994-497c-93f6-
                                                                    b2e913b07f59.docD06_DRAFT_V1
                                                                                6JW2 Andreas zweiter
                                                                         Teil.docD06_DRAFT_V15ejw




                      Wideband (better than PSTN)        < 100 ms
                      Narrowband (similar to PSTN)       < 150 ms
                      Best effort                        < 400 ms

                    Table 161616: Latency requirements
Note that the values in Table 16Table 16Table 16 refer to the total one-way roundtrip
latency including gateways, terminals, and jitter buffer. The delay in well-constructed
terminals/gateways for audio can be of the order 40-50 ms but can also be significantly
higher. In general, the latency is higher for video terminals than for audio terminals. The
latency is a statistical variable and a proposed requirement for the “wideband QoS class
reflecting this would be: 90% of the packets should have less than 40 ms latency.
Requirements on packet loss: Error concealment in mpeg-4 codecs (and packet loss
concealment in audio codecs) reduces the requirements on packet loss. Retransmission
cannot be used due to requirement on the latency. The situation is then similar to that of
IP-TV, approx. 0.5% maximum packet loss.
Coding “latency” – needs to be cosideredconsideredmmented for any video conference.
Typical MPEG-2 is “seconds” even for professional equipment at > 10kEuro, i.e. strictly
non-realtime. H.263 is “real-time” but operates typically at 10-15 fps and of low CIF
resolution. or lower.
Jitter is compensated by a jitter buffer at the receiving end. The requirements on the
latency limits the size of the jitter buffer to the order of 20 ms. Due to the demands on low
latency there is no time for retransmission similar to IP-TV. This is based on that algorithms
for packet loss concealment are used in the Mpeg-4 codecs.
Provisioning
Maximum set-up, teardown and minimum lease time are in the order of seconds for an
audio service and maybe a minute for a video service.
Dynamics
The bit-rate and the QoS parameters don’t change during the lease time. In conferences,
both audio and video, it should be possible to add and remove participants dynamically.
Connectivity – Transparency – Interface
VoIP and video chat can be either point-to-point or multi-point-to-multi-point applications all
carried by the IP layer.
                                                                                                     Formatted: Bullets and Numbering
       4.2.44.3.4 Video Conference
Resilience
Intrinsically, when an application becomes tool used for business, the resilience must rise.
Our opinion is that videoconference must have a high level of availability. For all services, a
99.99% of availability (53 min of downtime per year) may be a good level of availability.
Provisioning




                                       Page 53 of 168
                             IST IP NOBEL "Next generation
                                                                               Title of the document
                             Optical network for Broadband
                                  European Leadership"                    e1686d81-2994-497c-93f6-
                                                                   b2e913b07f59.docD06_DRAFT_V1
                                                                               6JW2 Andreas zweiter
                                                                        Teil.docD06_DRAFT_V15ejw




The lease time of a videoconference application can fluctuate from some minutes from
brief communications to hours for real audio-conferences. The set-up and teardown times
is better that are few seconds or less.
Dynamics
Once instituted the parameters related to the videoconference connection may remain
constant.
Bit-rate
The options and the choices are similar to the case of video chat, (and in addition to the
video streaming case there is a gain due to scene simplicity and a penalty due to real-time
compression). On the NOBEL timescale we assume mpeg-4 based codecs. However,
application that targets enterprise customers here will be a demand for higher video quality
then in the case of video chat. Services will range from equivalent to VHS quality and
upwardsupward.
Some trials executed in JUNIPER2 Eurescom project [P807] say that 200 kbit/s are
enough for a satisfactory Videoconference. The evolution of video quality encourages
having for this service larger bandwidth (about 500 kbit/s).
QoS
Latency requirements are according to the highest class VoIP: < 100ms. Packet loss
<0.5% and jitter <10 ms.
For video enterprise solutions, Video Conference in particular, quality is very important. In
detail, it is not so important to keep short latency (JUNIPER2 affirms that 1.2 sec is the
maximum even if the increase of the quality of the images shaping the video forces to
reduce this limit to 500 ms). More important is the issue of maximum difference of latency
that must remain less then 50 ms. Usually this application is carried by Layer-3; packet loss
must remain below 5% to have an acceptable quality. These limits are established by
interviewing users degrading gradually the quality parameters until the user says that the
service is not acceptable.




Connectivity – Transparency – Interface
Video Conference may be both point-to-point and multicast. For all the applications (ever
since they are addressed to a residential market) are carried on L3. It is possible a L2
connectivity for enterprise videoconference.
Security
Since this application may carry important information, it is desirable that the level of
security is high. In particular, AAA and cryptography may be useful.


                  Video application      Transport     Edge   Channel

                  on demand                   IP        AAA      Crypt.




                                      Page 54 of 168
                               IST IP NOBEL "Next generation
                                                                              Title of the document
                               Optical network for Broadband
                                    European Leadership"                 e1686d81-2994-497c-93f6-
                                                                  b2e913b07f59.docD06_DRAFT_V1
                                                                              6JW2 Andreas zweiter
                                                                       Teil.docD06_DRAFT_V15ejw




                   broadcast                   IP               Crypt.

                   download                    IP        AAA    Crypt.

                   Chat                        IP

                   Conference [ENT.]        IP-GbE       AAA    Crypt.

                  Table 171717: Security requirements for video applications
                                                                                                  Formatted: Bullets and Numbering
        4.2.54.3.5 Data applications (in particular video download)
Residential data applications means downloading files, which may have multimedia
content, or not. The experienced quality perceived by the end-user is determined by the
time that it takes to complete the download, so it is directly related to the bandwidth, the
more the better. From the user perspective, video download should be faster thaen real-
time (but it might be possible to download the video anyway). The usable bit-rate depends
on the terminal equipment, i.e. the network card, the memory and the disk space of the
computer, and it is probably not less thaen 100 Mbit/s. Other QoS parameters are less
important, 1s jitter, 5% packets loss and 20s latency might not compromise the success of
the application task.
The preferred service would generate very dynamic traffic patterns with point-to-point
bursts or short calls of high bit-rate (100 Mbit/s) followed by long periods of silence. The
maximum set-up and teardown are in the order of tens of seconds up to maybe a minute.
Minimum lease time is around a second. The maximum lease time depends on the bit-
rate,rate; if the largest data blocks are a couple few of Gbits the maximum lease time is
half a minute at 100 Mbit/s and 5 minutes at 10 Mbit/s.
                                                                                                  Formatted: Bullets and Numbering
        4.2.64.3.6 On-line gaming
On-line gaming is a rapidly growing business segment, in the US online gaming in the U.S.
alone will grow to approximately 24% of all gaming revenues (~$10.8B) in 2006 (Source:
Jupiter Media Metrix). Indeed, there could be implications to widespread online gaming for
ISPs, because even in 2002 about 9% of traffic across the U.S. Internet backbone involved
online gaming, reports In-StatMDR (second source needed). The usage of backbone
bandwidth is related to the huge numbers of gamers, each consuming in the order of
15-20 kbit/s. There are three main groups of online games:
    -   First person shooters (FPS) e.g. Counterstrike
    -   Massively multi-player online role playing game (MMORPG),
    -   Real time strategy (RTS) e.g. Warcraft III
There are two models of online gaming:
   Peer-to-peer model: players send their actions to each other and react on the received
    action. In peer-to-peer, each player communicates with every other player while state
    inconsistencies are resolved through a distributed agreement protocol.



                                       Page 55 of 168
                              IST IP NOBEL "Next generation
                                                                                 Title of the document
                              Optical network for Broadband
                                   European Leadership"                    e1686d81-2994-497c-93f6-
                                                                    b2e913b07f59.docD06_DRAFT_V1
                                                                                6JW2 Andreas zweiter
                                                                         Teil.docD06_DRAFT_V15ejw




   Client-server model: an authoritative game server is set up and all players or clients
    contact this game server to play the game against one another. Players send action
    messages to the game server. The game server processes the actions in sequence,
    changes the global state, and notifies players of the effects of their actions in state
    update messages. The communication is only between the game server and players. In
    the client-server model, the central server is responsible for resolving any state
    inconsistencies.
Today most of online games are implemented based on a client-server model due to the
complexity of peer-to-peer model implementation and security restrictions that prevent
peer-to-peer communications.
The main requirement on the network requirement is related to the latency (or ping = round
trip time). For fast action FPS games, a delay or ping below 50 ms is regarded associated
with excellent game play. A ping below 100 ms is good and above that, playability
decreases noticeably. Ping times above 150 ms are often reported intolerable. For RTS
games, e.g., the latency requirement is not strict and comparable to that of web browsing
(ranging from several hundreds of milliseconds to several seconds).
Bandwidth requirement is low, currently, for FPS games, typical data rate of each client or
player (and the data rate generated by the server for each individual player) are about 15-
20 kbit/s. In the future, the game traffic is not likely to change from what it looks like now,
except with the addition of voice communication, which will be incorporated in many online
games soon. In an investigation by In-Stat research 75% of the online gamers claimed that
they did not feel a need for a broadband connection or they though it was too expensive.
The main advantages with broadband access are the “always on” character of the
connection and the possibility of receiving incoming phone calls while playing.
The network service that would fit the requirements of gamers the best is a latency free
narrowband (i.e. cheap) connection.


       4.3.7 Mobile multi-media
The two major forces driving telecom today are Mobile and Internet. The classic voice
service is re-interpreted as mobile voice and voice-over-IP, and the original fixed voice
traffic is migrating into mobile and IP networks, with major economic impacts on operators
since the new networks often are owned and operated by other providers, and the revenue
streams are entirely different. Therefore, the classic voice is not likely to be a major driver
for new telecom network capacity or functionality in telecom-mature areas.
Simultaneously, both Mobile and Internet is moving into multi-service networks providing
much more than voice, and voice becomes an integral part of a richer set of
communication services. An integral part of this trend is that the mobile telephone is
becoming the ubiquitous device to always bring along; not only for voice, but also for your
important data, calendar, music and images. This service - all-in-one and always with you -
cannot be matched by fixed networks whichnetworks that will become focused on other
parameters (high bandwidth services). The requirements from the mobile end user will
therefore be focused on “always connected”, that is on high availability (any time), high
connectivity (anywhere) and for many different services (anything). This points to a high-
QoS, full multi-service network.




                                       Page 56 of 168
                              IST IP NOBEL "Next generation
                                                                                  Title of the document
                              Optical network for Broadband
                                   European Leadership"                     e1686d81-2994-497c-93f6-
                                                                     b2e913b07f59.docD06_DRAFT_V1
                                                                                 6JW2 Andreas zweiter
                                                                          Teil.docD06_DRAFT_V15ejw




As the air-interface will always limit the bandwidth of mobile services, the full bandwidth
services (also needing large screens and power supplies to be effective) will therefore
migrate towards fixed networks, possibly in combination with non-mobile “wireless drops”.
The requirements put on the transport network from the Mobile networks will therefore
probably focus on a high QoS, multi-layer, multi-service, transport network with modest
bandwidths. More important will be the support for efficient operations, fast roll-outrollout,
flexibility and availability that serves to build and maintain the network efficiently.
As mobile networks are operated by a variety of operators from incumbents to “virtual
operators”, they may want to lease transport capacity at any layer. Incumbents may build
and run their mobile networks directly as Layer 1 services (possibly on leased dark fibre),
whereas other operators may use leased lines (Layer 1). Most operators today are looking
at future cost-savings by utilizing fractions of existing transport capacity in Metro networks.
This would lead to a desire to operate with Layer 2 VPN services. Likewise, the data traffic
may be shifted into Layer 3 and be run over generic IP networks. This development is not
clear today, but in all cases, the orientation of the Mobile operator is likely to be to use a
fully layered, full-service, full-QoS network.
                                                                                                      Formatted: Bullets and Numbering
       4.2.74.3.8 Tele-medicine applications
Tele-medicine applications are a set of applications that can help medical acts required at
different steps of a medical care cycle. For instance, a cycle can start from with a
notification of an accident that has to be followed by its localisation, sending a rescue team,
diagnostics on the premises, transfer of medical data, transfer of the patient, admission to
a hospital, extended diagnostics – telediagnostics, and finally remote surgery or transfer of
the patient to a more appropriate hospital. These different steps include the following
application domains:
   Tele (remote) diagnostics
   Remote (tele) surgery
   Medical data storage
   Paramedic emergency communications

Note that communications are sometimes established between a moving vehicle and fixed
locations. This implies the use of mobile or satellite networks.
                                                                                                      Formatted: Bullets and Numbering
                      4.2.7.14.3.8.1 Tele- (or remote-) diagnostics
There are two types of diagnostics as a function of the time dependency: real time and off-
line.
The real time diagnostics are based on a real time observation of a patient by a medical
team located at a remote site with or without help of the local team (with the patient).
In a classical scheme, the local team executes all operations needed for establishing
accurate diagnostics and only provides the remote team with medical images and data
while discussing via videoconferencing facility.
In a more evolved scheme where a remote controlled robot is involved, the local team
typically sets up the robot at the patient (or the patient in the robot) and hands the control




                                        Page 57 of 168
                             IST IP NOBEL "Next generation
                                                                               Title of the document
                             Optical network for Broadband
                                  European Leadership"                    e1686d81-2994-497c-93f6-
                                                                   b2e913b07f59.docD06_DRAFT_V1
                                                                               6JW2 Andreas zweiter
                                                                        Teil.docD06_DRAFT_V15ejw




of the robot over to the remote team. The remote team then establishes diagnostics using
the robot. Typically, a videoconferencing is activated in-between for conversations and
visualisation of the global scene. At the same time, images and data synchronised to (or
captured by) the robot are transmitted to the robot operator. The robot can be equipped
with a feedback actuator. The use of a robot implies a short roundtrip delay (<500 ms, if
jitter <5 ms or <80 ms with jitter <40 ms) and a small synchronisation jitter among the
images, data, voice and robot controller (10 ms for haptics flows, a low impact of jitter on
images and voice). In addition, the feedback loop of the robot has to be very reliable (very
short disruption time, <40 ms). Therefore, those must be transported through a single
connection or a bundle of connections over the same route. 1+1 protection is necessary.
The minimum required bandwidth, depending on the definition and number of images per
second, would be several tens Mb/s in each direction (however it depends on application
requirements (for instance ranging from 1-10Mb/s) and flows may not be symmetrical –
high bandwidth for the patient site to remote (expert) site may be necessary.).
The off-line diagnostics mainly involve shared images and/or data for videoconferencing
between (or among) two (or several) sites. The shared images can be three-dimensional
objects. The requirements are similar to those of standard videoconferencing.
                                                                                                   Formatted: Bullets and Numbering
                      4.2.7.24.3.8.2 Remote (tele-) surgery
Remote surgery is similar to the real time diagnostics with a remote controlled robot except
for the required precisions in the operation. Here, the required precision is more severe
compared to real time diagnostics due to the impact on the patient. The roundtrip delay
(<200 ms) and synchronisation jitter must be carefully controlled. A reliable connection with
very small disruption time is necessary. The highest grade of end-to-end protection (1+1)
has to be used. Again, the minimum required bandwidth, depending on the definition and
number of images per second, would be several tens Mb/s or more in each direction.
Finally, the connection should not suffer from any external perturbations.
For instance, an experiment realised by France Telecom in 2001 between New York (USA)
and Strasbourg (France), 7000 km distant, used an ATM CBR connection at 10 Mb/s. This
bandwidth was divided into two traffic classes: Priority 1 and 2. Priority 1 traffic includes
robot control traffic (with a minimum guaranteed bit-rate of 512 kb/s), voicecal traffic over
IP telephony (with a minimum guaranteed bit-rate of 512 kb/s) and video signal of
endoscope with a minimum guaranteed bit-rate of 7 Mb/s. Video conference traffic is
carried by a reserved bit-rate of 2 Mb/s with Priority 2. The roundtrip delay was 152 ms,
divided into an 82 ms network delay and a 70 ms CODEC delay.
                                                                                                   Formatted: Bullets and Numbering
                      4.2.7.34.3.8.3 Medical data storage
Medical data storage is the same as other data storage applications. Its requirements stem
from security considerations for access.
                                                                                                   Formatted: Bullets and Numbering
                      4.2.7.44.3.8.4 Paramedic, emergency communications
This domain covers the localisation of patient, sending a rescue team, first diagnostics on
the premises, transfer of medical data, and transfer of the patient. This requires high
capacity wireless networks (similar to emergency ambulance vehicle to headquarter
communication networks) to transmit high definition images when necessary. The



                                      Page 58 of 168
                             IST IP NOBEL "Next generation
                                                                                Title of the document
                             Optical network for Broadband
                                  European Leadership"                    e1686d81-2994-497c-93f6-
                                                                   b2e913b07f59.docD06_DRAFT_V1
                                                                               6JW2 Andreas zweiter
                                                                        Teil.docD06_DRAFT_V15ejw




positioning systems are also required to guide the rescue team. The communications
should be carried by congestion free resilient networks that are videoconference capable.
The requirements should be similar to those of videoconferencing.
                                                                                                    Formatted: Bullets and Numbering
 4.34.4 Grids

       4.3.14.4.1 Definition
Currently several different definitions and views about GRIDsGrids – depending on
scenarios and services – are commonly used, but in any case, paves the way for building
virtual organizations. The term Grid is usually used in the scientific community to solve
problems, which are too big to be run on a single site. Hence, the geographical topology of
the Grid depends on the distribution of the community members. Though there might be a
strong relation between the entities building a virtual organizationorganisation, a Grid still
consists of resources owned by different, typically independent organizationsorganisations.
Heterogeneity of resources and policies is a fundamental result of this. Grid services and
applications therefore sometimes experience a quitequite different resource behaviour than
expected. Similarly, a distributed infrastructure with ambitious service demands puts stress
on the capabilities of the interconnecting network more than other environments. Grid
applications therefore often identify existing bootlenecks, either caused by conceptual or
implementation specific problems, or missing service capabilities.
The origin of the term Grid is an analogy to the term “Power Grid” which provides
“consistent, pervasive, dependable, and transparent access to resources”, which are
geographically dispersed. These resources are globally distributed including storage
systems, memory capacity, and computation power. For example, using Grid-based
systems, companies can create a “Virtual Enterprise” to share resources, which can be
added and removed as needed. Grids define different types of services, which are provided
to the end-user:
      Computational Grids
      Data Grids
      Utility Grids
However, in several scenarios, a combination of computational and data Grids is used in
order to process and store several 100 GbBytes and more of data per day. Depending on
size and locations, Grids can also be classified in cluster Grids, campus Grids, and global
Grids. A cluster Grid is usually located in one single location, without the need for long
distance connections while a campus Grid spreads over a campus of a single organisation.
The following sections will concentrate on global Grids only because of the nature of the
NOBEL project.
It must be noted that the Grid concept has several similarities with Peer-to-Peer (P2P)
applications but there are also differences (see Figure 9Figure 9Figure 9). P2P applications
are commonly used as a term for file sharing applications, which are run on a huge number
of standard personal computers. P2P in general refers also to compute intensive
applications, which are distributed to a huge number of systems (like seti@home). P2P
applications are usually not running under the control of a management entity and are only
providing a single service. The data volume transferred between the nodes is much smaller



                                       Page 59 of 168
                              IST IP NOBEL "Next generation
                                                                                Title of the document
                              Optical network for Broadband
                                   European Leadership"                   e1686d81-2994-497c-93f6-
                                                                   b2e913b07f59.docD06_DRAFT_V1
                                                                               6JW2 Andreas zweiter
                                                                        Teil.docD06_DRAFT_V15ejw




than in Grid applications and P2P nodes are not running in trusted environments. 1
Generally, peer-to-peer systems are distributed Internet applications in which the resources
of a large number of autonomous participants are harnessed in order to carry out the
system's function. In many cases, peers form self-organising networks that are layered
over the top of conventional Internet protocols and have no centralized structure.




            Figure 999: Main differences between P2P and GRID concepts
                                                                                                    Formatted: Bullets and Numbering
                      4.3.1.14.4.1.1 Compute Grids
These Grids are created solely for the purpose of providing access to computational
resources. Many applications today need computational resources, which cannot be
delivered by a single node or from a single location.
                                                                                                    Formatted: Bullets and Numbering
                      4.3.1.24.4.1.2 Data Grids
Grid deployments that require access to and processing of data are called data Grids. They
are optimised for data-oriented operations. Although they may consume a lot of storage
capacity, these Grids are not to be confused with general storage services.


1
 Most of the compute intensive P2P applications are following a server-based approach. In this
case a server or server farm sends small data packets to the P2P clients, which will do the
computation.




                                       Page 60 of 168
                            IST IP NOBEL "Next generation
                                                                             Title of the document
                            Optical network for Broadband
                                 European Leadership"                  e1686d81-2994-497c-93f6-
                                                                b2e913b07f59.docD06_DRAFT_V1
                                                                            6JW2 Andreas zweiter
                                                                     Teil.docD06_DRAFT_V15ejw



                                                                                                 Formatted: Bullets and Numbering
                     4.3.1.34.4.1.3 Utility Grids
We define utility Grids as commercial compute resources that are maintained and
managed by a service provider. Customers that have the need to augment their existing,
internal computational resources may purchase “cycles” from a utility Grid. In addition to
overflow applications, customers may choose to use utility Grids for business continuity
and disaster recovery purposes.
                                                                                                 Formatted: Bullets and Numbering
       4.3.24.4.2 Service Requirements
Currently several Grid applications are requiring high bandwidth connections over large
distances with small delays and QoS. The following section will give to examples in order
to understand the transport services, which need to be provided by the underlying network
infrastructure.
                                                                                                 Formatted: Bullets and Numbering
                     4.3.2.14.4.2.1 Grid Scenarios
HEP (High Energy Particle Physics):
HEP is at the moment the most demanding application for Grid environments. The data
rate of a Large Hadron Collider like Cern currently leads to a few Petabytes a year;
estimations for 2012 result in data rates in the range of several Exabytes/year. Today’s
data sets have a size between 10 and 100 Terabyte, which takes up to 24 hours to transmit
over a 10Gbit/s link. Depending on the state of the LHC high bandwidth is needed on
demand when data is created by the LHC. Due to the huge amount of data the analyses
and computation can’t be done locally but has to be performed on locations which are
geographically wide spread (up to several 1000 km).




                                     Page 61 of 168
                              IST IP NOBEL "Next generation
                                                                                  Title of the document
                              Optical network for Broadband
                                   European Leadership"                     e1686d81-2994-497c-93f6-
                                                                     b2e913b07f59.docD06_DRAFT_V1
                                                                                 6JW2 Andreas zweiter
                                                                          Teil.docD06_DRAFT_V15ejw




Figure 101010: Example of Multi Tier structure
Currently a multi-tier structure is used to interconnect the different locations. The demand
for the bandwidth is expected to grow up to 40 GbBit/s per Tier 0 link. Figure 10Figure
10Figure 10 shows the multi tier structure used in the Cern Grid. The structure follows a
tree with decreasing bandwidth demands as the distance from the root node is growing.
Depending on the scenarios it is also likely that the topology of the Grid is going to change;
data will not only flow with decreasing bandwidth down the existing tree but also between
Tier1 and Tier2 nodes with several GBitGbit/s. The underlying network must be able to
provide a flexible resource allocation mechanisms whichmechanisms that can be
controlled using the Grid rResource bBroker.
High Performance Visualisation:
Grids allow the separation of computation and the visualisation of computed results. In this
case the computation is done on one or more high performance compute cluster and the
results can be visualized on remote machines in the Grid. End users at the terminals can
interact with the simulation in real time. This results in data flows of about 1 GBitGbit/s with
low delay and small jitters. As an option multicasting can be used if more than one
observer is watching the same simulation.
Astrophysics:
Radio astronomers are using Very Long Baseline Interferometry (VLBI) to collect detailed
data from cosmic radio sources. For a higher resolution several geographically distributed
radio telescopes are coordinated to build a single telescope. In the standard approach the
data is collected locally with high-resolution time stamps (e.g. on tapes). The tapes are
shipped to a central site where the correlation is performed. This will lead to long delays
until the final results are available.



                                       Page 62 of 168
                              IST IP NOBEL "Next generation
                                                                                                                     Title of the document
                              Optical network for Broadband
                                   European Leadership"                                               e1686d81-2994-497c-93f6-
                                                                                               b2e913b07f59.docD06_DRAFT_V1
                                                                                                           6JW2 Andreas zweiter
                                                                                                    Teil.docD06_DRAFT_V15ejw




                          Figure 111111: European VLBI Network
Using high speed Grids it is possible to correlate the data streams from all radio telescopes
in the VLBI in real time at a central site. In this scenario every node can create data rates
of about 1GBit/s (with future requirements up to 40GBit/s).
                                                                                                                                         Formatted: Bullets and Numbering
       4.3.34.4.3 Parameters

                      4.3.3.14.4.3.1 Resilience
It is quite difficult to define resilience requirements for Grids, because they will depend on
the application running over the Grid. However, we can extrapolate the following issues.
Regarding to the availability parameter, it will take into account free Grid resources in each
moment. Therefore, it will depend on the instant that we are facing.
Applications for this technology may cater to time-sensitive applications by offering an
intelligent protection switching, which “self-heals” a network in the event of a fibre fault,
promising to restore the circuit within 50ms. Experiments have been conducted to support
this claim.


                                                                                                                                         Formatted: Bullets and Numbering
                                                               Max recovery time




                      1
                                                                                                   Max outage time
                                            Min Availability




                                                                                   Block prob of




                       Application
                                                                                   restoration
                                                               [ms]




               Compute Grid                4-      50                                              hours
                                           90,9999




                                       Page 63 of 168
                                 IST IP NOBEL "Next generation
                                                                                                                                                       Title of the document
                                 Optical network for Broadband
                                      European Leadership"                                                                     e1686d81-2994-497c-93f6-
                                                                                                                        b2e913b07f59.docD06_DRAFT_V1
                                                                                                                                    6JW2 Andreas zweiter
                                                                                                                             Teil.docD06_DRAFT_V15ejw




              Data Grid                        0,99994 50                                                                hours
                                               -9
              Utility Grid                     0,99999 50                                                                minutes
                                               5-9

                  Table 181818: Some QoS requirements for GRID
                                                                                                                                                                            Formatted: Bullets and Numbering
                      4.3.3.24.4.3.2 Provisioning
For Grid solutions, the provisioning time (set-up and tear-down) should be as shortest as
possible. Moreover, in this kind of applications, it is obvious that a very large amount of
data could be processed. Therefore, the lease time would normally fluctuate between
hours and days,days as shown in see Table 19Table 19Table 19.
                                                                      Max tear down time
                                                 Max set-up time




                                                                                                                                     Max lease time
                                                                                                       Min lease time




                                                                                                                                                             Max blocking
                             Application




                 Compute Grid                    sec                  sec                           hours                          days                      probability

                 Data Grid                       sec                  sec                           hours                          days
                 Utility Grid                    sec                  Sec                           minutes                        days

                Table 191919: Provisioning requirements for GRID
                                                                                                                                                                            Formatted: Bullets and Numbering
                      4.3.3.34.4.3.3 Dynamics
It seems important to stress that in a Grid application it will be necessary to manage its
parameters dynamically. We should remember that a Grid is a collection of distributed
computing resources. We are going to use only those resources available in each moment,
so we will need to assume changes in bitrate, in QoS and in connectivity.
                                                 Bitrate management




                                                                                                                                   Connectivity mgmt
                                                                                           QoS management




                             Application



                                                                                                                                                                            Comment [TZ2]: Eine Null-Information !


                 Compute Grid                    Y                                         Y                                     Y




                                           Page 64 of 168
                                IST IP NOBEL "Next generation
                                                                                                                                                       Title of the document
                                Optical network for Broadband
                                     European Leadership"                                                                          e1686d81-2994-497c-93f6-
                                                                                                                            b2e913b07f59.docD06_DRAFT_V1
                                                                                                                                        6JW2 Andreas zweiter
                                                                                                                                 Teil.docD06_DRAFT_V15ejw




                 Data Grid                     Y                                                              Y                                    Y
                 Utility Grid                  Y                                                              Y                                    Y

Table 202020: Dynamics requirements for GRID
                                                                                                                                                                           Formatted: Bullets and Numbering
                       4.3.3.44.4.3.4 Bit-Rate
To cope with the bit-rate problem, there are some points to be cleared up. First, in a Grid
application there will be numerous connections, not only from the users to the Grid, but
also inside of the Grid. Nominal aggregated bit-rate will be strongly associated with the
application that is carried out. Besides, it will also depend on the network implementation.
Therefore, it is quite difficult to talk about a maximum rate, and the minimum might be
imperceptible. This fact will lead to a wide range, difficult to be measured.
                                               Max nominal bitrate

                                                                     Min Dynamic range


                                                                                         Min Increment time


                                                                                                                  Max Decrement


                                                                                                                                  BR Granularity
                          Application
                                                                                                                  time




                 Compute Grid                                                            sec                      sec             Kbps
                 Data Grid                                                               min                      min             Mbps
                 Utility Grid                                                            sec                      sec             Kbps

                Table 212121: Bit-rate requirements for GRID
                                                                                                                                                                           Formatted: Bullets and Numbering
                       4.3.3.54.4.3.5 QoS
A comprehensive architecture for providing applications with Quality of Service for different
types of resources would have to:
      Provide a flexible architecture that can provide QoS for different types of resources,
       including networks, CPUs, batch job schedulers, disks, and graphic pipelines.
      Provide mechanisms to allow both advance reservations and immediate
       reservations for quality of service. An immediate reservation begins right after
       creating it, whereas an advance reservation begins at some future time.
   Enable high-performance computing users to conveniently make and use QoS
      reservations for complex sets of resources. For example, a scientific application
      may want to request nodes on a supercomputer, guaranteed access to disks on
      that supercomputer, real-time scheduling on a remote visualization computer, and
      guaranteed network bandwidth between the two.
   
Reservations for QoS have several interesting features:




                                        Page 65 of 168
                               IST IP NOBEL "Next generation
                                                                                                                                                     Title of the document
                               Optical network for Broadband
                                    European Leadership"                                                                        e1686d81-2994-497c-93f6-
                                                                                                                         b2e913b07f59.docD06_DRAFT_V1
                                                                                                                                     6JW2 Andreas zweiter
                                                                                                                              Teil.docD06_DRAFT_V15ejw




      Reservations are uniform, first-class entities.
      Reservations can be created, modified, and cancelled.
      Reservations can be monitored, either through a poling mechanism, or through
       callbackcall-back functions.


Although the reservations appear uniform to the user, the underlying implementations are
not. It is flexible enough to easily interface with various reservation systems.
In order to fill the following table, we will split the compute Grid type into two classes. The
first class comprises those applications that need almost a real-time interaction between
the main system and the Grid. The second one is comprised of applications that are more
flexible and no real-time interaction is needed.


                                                                                                                           Packet Loss (Layer 3)
                                            Max latency time [ms]



                                                                        Max diff Latency [ms]



                                                                                                Max BER (Layer 1+)




                       Application




              Compute Grid
                       Real time interaction 100                        20                                                 0,01%
                   Non real time interaction 500                        100                                                0,1%
              Data Grid                     500                         100                                                0,1%
              Utility Grid                  200                         50                                                 0,01%

             Table 222222: QoS requirements for GRID
                                                                                                                                                                         Formatted: Bullets and Numbering
                        4.3.3.64.4.3.6 Connectivity – Transparency – Interface
Connectivity seems to be inferred from the concept of Grid applications. Since a Grid is a
collection of distributed computing resources, connectivity in this kind of applications will be
N:M in all-different cases.
Besides, all the applications are carried on Layer 3 over an IP interface.


                                            Connectivity Transparency                                                                              Interface
                                                                                                                                                                         Formatted: Bullets and Numbering

                          Application                               1                                                1                                  1




                                        Page 66 of 168
                                 IST IP NOBEL "Next generation
                                                                                 Title of the document
                                 Optical network for Broadband
                                      European Leadership"                 e1686d81-2994-497c-93f6-
                                                                    b2e913b07f59.docD06_DRAFT_V1
                                                                                6JW2 Andreas zweiter
                                                                         Teil.docD06_DRAFT_V15ejw




                  Compute Grid              N:M            L3               IP
                  Data Grid                 N:M            L3               IP
                  Utility Grid              N:M            L3               IP

           Table 232323: Connectivity, transparency, and interface requirements for
GRID
                                                                                                     Formatted: Bullets and Numbering
                       4.3.3.74.4.3.7 Security
Grid Computing covers a broad spectrum of interconnected computer systems, ranging
from tightly managed supercomputers to more independently used desktops and laptops
distributed over a wide area network. The security problem in a Grid environment is
complex because resources are often located in different administrative domains with each
resource potential having its own policies and procedures. Despite the definition and
observation of strict security policies, Grids are still vulnerable to security attacks at all
network points, with the greatest threat remaining at the desktop level. Some Grid
frameworks offer end-to-end security that includes sand boxing of jobs during execution on
compute nodes. The sand-boxing feature enforces a security wall between the job
execution environment and the compute node on which it is running. For examples, a
desktop user will be unable to view or manipulate the Grid job running on his computer.
Therefore, it seems clear that security level must remain high.
Authentication, confidentiality, data integrity and non-repudiation are some basic features
required to operate a secure Grid.
More in detail:
      Authentication - A method for verifying that the sender of a message is really who
       he or she claims to be. Any intruder masquerading as someone else is detected by
       authentication.
      Data iIntegrity (checking) - A method for verifying that a message has not been
       altered along the communication path. Any tampered message sent by an intruder
       is detected by an integrity check. As a side effect, communication errors are also
       detected.
      Non-repudiation - The possibility to prove that the sender has really sent the
       message. When algorithms providing non-repudiation are used, the sender is not
       able to later deny the fact that he or she sent the message in question.


                      Grid application Transport Edge Channel

                      Compute Grid                IP      AAA    Crypt
                      Data Grid                   IP      AAA    Crypt
                      Utility Grid                IP      AAA    Crypt

                     Table 242424: Some security requirements for GRID




                                         Page 67 of 168
                                             IST IP NOBEL "Next generation
                                                                                                                                                                                                                                    Title of the document
                                             Optical network for Broadband
                                                  European Leadership"                                                                                                    e1686d81-2994-497c-93f6-
                                                                                                                                                                   b2e913b07f59.docD06_DRAFT_V1
                                                                                                                                                                               6JW2 Andreas zweiter
                                                                                                                                                                        Teil.docD06_DRAFT_V15ejw



                                                                                                                                                                                                                                                                                                          Formatted: Bullets and Numbering
 4.44.5 Summary of emerging applications
The table shows a summary of the parameters required by the applications described in
the current chapter.
It is important to note that the most important emerging applications are considered in the
current study and, in particular, even if the parameters related to some secondary
applications are not directly esteemed, the analysed applications cover a huge range of
parameters, so to analyse a complex and variegated scenario that in all probability it is not
far from the evolutions of the applications requirement in the next future.


                        User Application                                   Resilience                                                                        Provisioning                                                                          Dynamics
                                                                                      Max recovery time [ms]




                                                                                                                                                                                                                                        Bitrate management
                                                                                                                                                                   Max tear-down time




                                                                                                                                                                                                                                                                                      Connectivity mgmt
                                                                                                                                                                                                                                                                 QoS management
                                                                                                                   Max outage time


                                                                                                                                           Max set-up time




                                                                                                                                                                                                                   Max lease time
                                                                  Min availability




                                                                                                                                                                                              Min lease time


 Storage
 - Back-Up / Restore                                             4-9                   0                       1 day                 min                     min                        min    hours N                                                       N                    N
 - Storage on demand (SoD)                                       5-9                 100                       hours                 sec                     sec                             0 perm. Y                                                       Y                    N
 - Asyncrhonous mirroring                                        5-9                 100                       hours                 sec                     sec                             0 perm. Y                                                       Y                    N
 - Synchronous mirroring                                         5-9                   0                       hours                 min                     min                        hours perm. N                                                        N                    N
 Grid computing
 - Compute Grid                                                  4_9                  50 hours                                       sec                     sec                        hours days                                  Y                        Y                    Y
 - Data Grid                                                     4_9                  50 hours                                       sec                     sec                        hours days                                  Y                        Y                    Y
 - Utility Grid                                                  5_9                  50 hours                                       sec                     sec                        minutesdays                                 Y                        Y                    Y
 Multimedia
 - Video on Demand (entertainment quality)                      99,5                 100                       hours                 sec                     sec                        min                    hours                N                        N                    N
 - Video Broadcast (IP-TV, entertainmant quality)               99,5                 100                       hours                 sec                     sec                        min                    hours                N                        N                    N
 - Video Download                                                4-9                 100                       hours                 sec                     sec                        min                    hours                N                        N                    N
 - Video Chat (SIF quality, no real-time coding penalty)        99,5                 100                       hours                 sec                     sec                        min                    hours                N                        N                    N
 - Narrowband Voice, data (VoIP,...)                             4-9                 100                       hours                 ms                      ms                         sec                    hours                N                        N                    N
 - Telemedicine (disgnostic)                                    5_9                   40                       hours                 ms                      ms                         sec                    hours                N                        N                    N
 - Gaming                                                       99.5                 100                       hours                 sec                     sec                        min                    hours                N                        N                    N
 - Digital distribution, digital cinema                          4-9                 100                       hours                 sec                     sec                        min                    hours                N                        N                    N
 - Video conference (PAL broadcast quality)                      4-9                 100                       hours                                                                    min                    hours                ?                        ?                    ?




                                                           Page 68 of 168
            IST IP NOBEL "Next generation
                                                           Title of the document
            Optical network for Broadband
                 European Leadership"                e1686d81-2994-497c-93f6-
                                              b2e913b07f59.docD06_DRAFT_V1
                                                          6JW2 Andreas zweiter
                                                   Teil.docD06_DRAFT_V15ejw




Table 252525: Application properties and parameters (1).




                    Page 69 of 168
                                                IST IP NOBEL "Next generation
                                                                                                                                                                                                                                          Title of the document
                                                Optical network for Broadband
                                                     European Leadership"                                                                                                               e1686d81-2994-497c-93f6-
                                                                                                                                                                                 b2e913b07f59.docD06_DRAFT_V1
                                                                                                                                                                                             6JW2 Andreas zweiter
                                                                                                                                                                                      Teil.docD06_DRAFT_V15ejw




                       User Application                                                         Bitrate [Mb/s]                                                                                                                                      QoS




                                                                                                                                                                                                                                                                                                Packet Loss (Layer 3)
                                                                                                                                                                                                                                          Max diff. Latency [ms]
                                                                                                                                                            Max Decrement time




                                                                                                                                                                                                                                                                         Max BER (Layer 1+)
                                                                       Max Nominal Bitrate




                                                                                                                                Max Increment time
                                                                                                     Min Dynamic range




                                                                                                                                                                                                                Max latency [ms]
                                                                                                                                                                                        BR Granularity
Storage
- Back-Up / Restore                                                   400 const                                          N.A.                        N.A.                                                N.A.                      N.A.                                                       0,1%
- Storage on demand (SoD)                                            1000                                                sec                         sec                                                     10                            1 1E-12                                            0,1%
- Asyncrhonous mirroring                                              400                                                sec                         sec                                                    100                           10 N.A.                                             0,1%
- Synchronous mirroring                                              2000 const                                          N.A.                        N.A.                                                     3                            1 1E-12
Grid computing
- Compute Grid                                             N.A.                              N.A.                        sec                         Sec                         Kbps                       100                        20                                                     0,0%
- Data Grid                                                N.A.                              N.A.                        min                         Min                         Mbps                       500                       100                                                     0,1%
- Utility Grid                                             N.A.                              N.A.                        sec                         sec                         Kbps                       200                        50                                                     0,0%
Multimedia
- Video on Demand (entertainment quality)                  2.5-5.0                           const                       N.A.                        N.A.                                                2-20s       50                                            N.A                        0,5%
- Video Broadcast (IP-TV, entertainmant quality)           2.5-5.0                           const                       N.A.                        N.A.                                                2-20s       50                                            N.A                        0,5%
- Video Download                                                       20                            19                                                                                                  2-20s    1000                                             N.A                        1,0%
- Video Chat (SIF quality, no real-time coding penalty)    0.2-0.4                                   0,5                                                                                                     400     10                                            N.A                        5,0%
- Narrowband Voice, data (VoIP,...)                        5,3-64 kb                         N.A                         N.A.                        N.A.                                                100-400     10                                            N.A                        0,5%
- Telemedicine (disgnostic)                                1-10                              N.A.                        N.A.                        N.A.                                                40-250 5-40                                               N.A                        0,5%
- Gaming                                                   10-25 kb                                  0,5                                                                                                 50-75       10                                            N.A                        5,0%
- Digital distribution, digital cinema                             1000                      const                       N.A.                        N.A.                                                    120     80                                            N.A                        0,5%
- Video conference (PAL broadcast quality)                 1.5-3.0                           const                       N.A.                        N.A.                                                    100     10                                            N.A                        0,5%




                        Table 262626: Application properties and parameters (2).




                                                          Page 70 of 168
                                  IST IP NOBEL "Next generation
                                                                                                                    Title of the document
                                  Optical network for Broadband
                                       European Leadership"                                                 e1686d81-2994-497c-93f6-
                                                                                                     b2e913b07f59.docD06_DRAFT_V1
                                                                                                                 6JW2 Andreas zweiter
                                                                                                          Teil.docD06_DRAFT_V15ejw




                       User Application                                          Connectivity   Transparency        Interface




                                                          Max Nominal Bitrate
Storage
- Back-Up / Restore                                                             p2p             L1-L2-L3       FC/IP/GbE
- Storage on demand (SoD)                                                       p2p             L1-L2-L3       FC/IP/GbE
- Asyncrhonous mirroring                                                        p2p             L3             IP/GbE
- Synchronous mirroring                                                         p2p             L1             FC/WDM
Grid computing
- Compute Grid                                                                  N:M
                                                                                N.A.            L3             IP
- Data Grid                                                                     N:M
                                                                                N.A.            L3             IP
- Utility Grid                                                                  N:M
                                                                                N.A.            L3             IP
Multimedia
- Video on Demand (entertainment quality)                                       p2p
                                                                                2.5-5.0         L3             IP
- Video Broadcast (IP-TV, entertainmant quality)                                multi-broad
                                                                                2.5-5.0         L3             IP
- Video Download                                                                p2p             L3             IP
- Video Chat (SIF quality, no real-time coding penalty)                         p2p
                                                                                0.2-0.4         L3             IP
- Narrowband Voice, data (VoIP,...)                                             p2p
                                                                                5,3-64 kb       L3             IP
- Telemedicine (disgnostic)                                                     p2p
                                                                                1-10            L1-L2-L3       IP/GbE
- Gaming                                                                        p2p
                                                                                10-25 kb        L3             IP
- Digital distribution, digital cinema                                          p2p-multi       L3             IP
- Video conference (PAL broadcast quality)                                      p2p-multi
                                                                                1.5-3.0         L3             IP/GbE




               Table 272727: Application properties and parameters (3).




                                               Page 71 of 168
                              IST IP NOBEL "Next generation
                                                                              Title of the document
                              Optical network for Broadband
                                   European Leadership"                  e1686d81-2994-497c-93f6-
                                                                  b2e913b07f59.docD06_DRAFT_V1
                                                                              6JW2 Andreas zweiter
                                                                       Teil.docD06_DRAFT_V15ejw




 5        Network Services
This section describes the five assigned groups of network services: their features,
relations, and differences. Based on customer category and needs, the grouping was made
according to: Public IP, Business IP, and Virtual Private Networks (VPN) on L3, VPN on
L2, and VPN on layer L1. The characteristics of each group are described along with their
performance parameters.
The different layers’ VPNs are explained from a perspective of connectivity, control,
scalability and flexibility. Mechanisms enabling the network “privacy” are described, such
as MPLS and tunnelling in packet networks, and lightwave services enabled by some
future L1-VPN. Finally there is considered how network services should match the
applications emphasized in chapter 4: storage, grid, and multimedia.


 5.1      Introduction
This section describes the five assigned groups of network services: their features,
relations, and differences. Based on customer category and needs, the grouping was made
according to:
        Public IP,                                                                               Formatted: Bullets and Numbering

        Business IP,
        Virtual Private Networks (VPN) on L3,
        VPN on L2, and
        VPN on layer L1.
The characteristics of each group are described along with their performance parameters.
The different layers’ VPNs are explained from a perspective of connectivity, control,
scalability and flexibility. Mechanisms enabling the network “privacy” are described, such
as MPLS and tunnelling in packet networks, and lightwave services enabled by some
future L1-VPN. Finally there is considered how network services should match the
applications emphasized in chapter 4: storage, grid, and multimedia.
First of all the network should allow communication between clients when they choose to
do so. A suitableSuitable connection availability must be offered. The request for an
information exchange should be allowed and given with a reasonable response time
(delay) and the information should be delivered with the required quality such as bit rate
and error rate, capacity and packet loss. The information rate and its variations will of
course also relate to the type of application running. The network should be available in
spite of planned outages, or problems due to failures. This might require either restoration
or protection switching to an alternative path using dedicated equipment. Protection
switching can be done in every layer, although the number of layers involved in the
decision should be kept at a minimum.
The aim for a network service would be to satisfy the requirements that a particular
application needs. The service must also be offered at an acceptable cost per bit.




                                      Page 72 of 168
                              IST IP NOBEL "Next generation
                                                                                Title of the document
                              Optical network for Broadband
                                   European Leadership"                   e1686d81-2994-497c-93f6-
                                                                   b2e913b07f59.docD06_DRAFT_V1
                                                                               6JW2 Andreas zweiter
                                                                        Teil.docD06_DRAFT_V15ejw




The packet network solution is very flexible to variations in capacity demand and the
amount of users due to the inherent statistics of packets. On the other hand, the packet
network will in general need a good balance between the number of priority switching
classes and the amount of administration due to the number of classes. The more classes
the higher network load can be tolerated. Networks based on IP, Ethernet, Optical Burst
Switching (OBS) and Optical Packet Switching (OPS) are packet solutions.
The connection oriented network approach will generally minimize the latency problem, but
sets a limit on the number of simultaneous “users”, i.e. connections. A connection oriented
network approach could be SDH, OTN and Lambda channel.


 5.2     Network service groups
                                                                                                    Formatted: Bullets and Numbering
       5.2.1 Introduction
Five groups of network service are assumed to be sufficient: Public IP, Business IP, and
Virtual Private Networks (VPN) on transport layer 3, VPN on layer 2, and VPN on layer 1.
Each of these services can be further classified, according to Network Service Table
29Table 29Table 29.
The IP network service is split into Public IP and Business IP, where the Public IP will be a
best effort service, and the Business IP will be a priority class that for example will handle
latency sensitive traffic. Business IP also is assumed to offer higher throughput than Public
IP.
Virtual Private Networks (VPN) on layer 3 carried over IP with an MPLS label will be
classified into a permanent configured network class and a network class that can be set
up on demand.
The VPN on layer 2 is carried either in a VLAN with an MPLS label, by Data over
Sonet/SDH over Ethernet, or bperhaps in future by Optical Burst Switching (OBS) or
Optical Packet Switching (OPS). Layer 2 VPN will have four classes corresponding to: if it
can be configured on demand or not, and if it has built-in protection switching or not.
The VPN on layer 1 will be classified like the VPN on layer 2, but carried either in SDH,
ODU, or in a transparent optical network with optical cross-connects or optical add drops
(e.g. wavelength service).
Before discussing the VPN services some layer-dependent features for VPNs on layer 3, 2
and 1 will be reviewed.




                                       Page 73 of 168
                            IST IP NOBEL "Next generation
                                                                            Title of the document
                            Optical network for Broadband
                                 European Leadership"                  e1686d81-2994-497c-93f6-
                                                                b2e913b07f59.docD06_DRAFT_V1
                                                                            6JW2 Andreas zweiter
                                                                     Teil.docD06_DRAFT_V15ejw




                    Figure 1212: Network architecture alternatives.
                                                                                                Formatted: Bullets and Numbering
       5.2.15.2.2 L3 VPN
In L3 VPN, L3 connectivity (normally IP) is used between the customer sites. The VPN is a
number of nodes (sites) that are connected by virtual links. Forwarding in the L3 VPN is
based on L3 information, e.g. IP addresses, and the provider edge (PE) and customer
edge (CE) exchange routing information, see Figure 12Figure 13Figure 13.




            Figure 121313: IP/MPLS based tunnel at Layer 3 between PEs.
The most important L3 VPN approach is the BGP/MPLS VPN solution. It concentrates the
VPN functionality to the provider edges (PE). By doing this it hides the VPN specific
information from the provider core nodes, thus improving scalability
.

In BGP/MPLS VPNs the PEs use MPLS labels to identify the VPNs and keep the VPN
traffic isolated from other VPNs. The packets are transmitted across the core network in
tunnels, which can be MPLS, IPSec, or Generic Routing Encapsulation (GRE) tunnels.
Certain procedures are common to both the MPLS-in-IP and the MPLS-in-GRE
encapsulations.
MPLS is most commonly used and, in this case, the transport tunnel is identified with
another MPLS label, see Figure 13Figure 14Figure 14.



                                     Page 74 of 168
                             IST IP NOBEL "Next generation
                                                                              Title of the document
                             Optical network for Broadband
                                  European Leadership"                   e1686d81-2994-497c-93f6-
                                                                  b2e913b07f59.docD06_DRAFT_V1
                                                                              6JW2 Andreas zweiter
                                                                       Teil.docD06_DRAFT_V15ejw




       Figure 131414: Using MPLS LSP when possible and MPLS IP otherwise.
This transport label is only significant between two nodes and is changed at every node.
The advertisement of the routes in the VPN is signalled between the PEs using BGP,
hence the name BPG/MPLS VPNs.
As the different VPNs can use private addresses (that can overlap) BGP has been
extended with a new address family group, the VPN-IPv4 family. In this, the route identifier
consists of both an IP prefix and a route distinguisher (RD). By keeping the RD unique for
each VPN it is possible to have overlapping address spaces in VPNs. IPv6 VPNs may thus
handle routing between different VPNs using the same address without conflicts, by using
the mechanism of PE translation of the 8 byte RD to the 24-byte VPN-Ipv6 address.
                                                                                                  Formatted: Bullets and Numbering
       5.2.25.2.3 L2 VPN
In L2 VPN, forwarding decisions in the provider network are based on L2 information only
such as MAC address, ATM VC identifier, MPLS label, and port number. (Figure 30) No L3            Comment [TZ3]: Welches Fig?
specific VPN information is used here.
L2 VPN gives the customer flexibility in terms of routing control and L3 protocol usage, as
the customer and the provider do not exchange any routing information with each other.
Ethernet, as data plane convergence layer controlled as L2 LSPs (defined in GMPLS
Architecture) for Metro Ethernet Networks, is a cost effective solution that simplifies
integration of native Ethernet services.
On the other hand, the Pseudo Wire Emulation Edge-to-Edge (PWE3) delivers, by
definition, only point-to-point services, neither P2MP, nor MP2P, nor MP2MP capabilities.
The Pseudo Wire functions include encapsulating (or tunnelling) the traffic, carrying them
across a path or tunnel, managing their timing and order, and any other operations
required to emulate the behaviour and characteristics of the service as faithfully as
possible. It is also called Circuit Emulation Services over Packet Switched Network (PSN).




                                      Page 75 of 168
                             IST IP NOBEL "Next generation
                                                                               Title of the document
                             Optical network for Broadband
                                  European Leadership"                   e1686d81-2994-497c-93f6-
                                                                  b2e913b07f59.docD06_DRAFT_V1
                                                                              6JW2 Andreas zweiter
                                                                       Teil.docD06_DRAFT_V15ejw




                 Figure 141515: L2 tunnelling principle based on PSN.
The PWE3 tunnelling principle may be described as follows:
      Transporting L2 PDUs from ingress PE router to egress PE router, across an inter-
       vening MPLS network over an LSP.
      Ingress PE router can cause a packet to be delivered to egress PE router by
       pushing some label onto the packet: “Tunnel label” (tunnel LSP).
      The tunnel LSP merely gets packets from ingress to egress router. The
       corresponding label doesn't tell what to do with the payload.
      Thus if the payload is not an IP packet, there must be a label, which becomes
       visible, that tells egress PE router how to treat the received packet. Call this label
       the “VC label”, Figure 15Figure 16Figure 16.                                                Formatted: Font: Arial

The PWE3 tunnelling implies that MPLS becomes the convergence layer as it was
intended through ATM.
Furthermore, it must be noted that using PWE3 with MPLS as intra-domain solution, the
LDP signalling requires pre-provisioning of end-to-end control channels, in addition to the
support of MPLS control plane within the network i.e. RSVP-TE signalling, rising thereby
the n2 scalability issue. As inter-domain solution, LDP signalling re-invents end-to-end
GMPLS RSVP-TE capabilities in addition to be non-scalable compared to the “flooding”
mechanisms via (multi-layer) routing.




       Figure 151616: PWE encapsulation with ‘demux’ and ‘switching’ labels.
On the other hand, in MPLS, the label implicitly refers to IP packet: An LSP is created, but
without precisely giving the type of the LSP involved, while in GMPLS the type of the LSP
is specified: Switching Capability (SC) and payload (GPID) Figure 16Figure 17Figure 17. It
is true even for an IP LSP. From the point of view of tunnelling GMPLS and PWE3 are
similar.




                                      Page 76 of 168
                             IST IP NOBEL "Next generation
                                                                              Title of the document
                             Optical network for Broadband
                                  European Leadership"                   e1686d81-2994-497c-93f6-
                                                                  b2e913b07f59.docD06_DRAFT_V1
                                                                              6JW2 Andreas zweiter
                                                                       Teil.docD06_DRAFT_V15ejw




     Figure 161717: GMPLS encapsulation with ‘payload’ and ‘switching’ labels.
However, how do the two different control plane cooperate?
In the first case, noted (1) in Figure 17Figure 18Figure 18, two control plane instances,
namely PWE3 for the L2 and GMPLS for the transport layer, should collaborate with an
accurate interlayer mechanism (PWE3 uses VC and TU labels).
The second scenario (2) corresponds to the case where the management plane performs
the generic control plane coordination in order to harmonize the operations of both the
PWE3 and GMPLS control plane instances.
The third case (3) describes a unique GMPLS control plane instance for both L2 tunneling
and lower transport layers that corresponds to the future full-integrated solution. In (3)
GMPLS operates as a unique, vertically integrated control plane providing a common
unified mean for all the switching capabilities present in the network, where, for instance,
the GMPLS Routing Controller maintains L2SC and TDM-SC topology and reachability
information within a single routing instance.




         Figure 171818: Migration towards a common control plane instance.
If instead the customer would like the VPN to emulate a LAN, VPLS (Virtual Private LAN
Service) can be used. However, the standardization efforts has not yet converged totally
for VPLS which means that the VPLS solution from one vendor may not be interoperable
with one from another vendor. [Rosenbaum]




                                      Page 77 of 168
                             IST IP NOBEL "Next generation
                                                                              Title of the document
                             Optical network for Broadband
                                  European Leadership"                   e1686d81-2994-497c-93f6-
                                                                  b2e913b07f59.docD06_DRAFT_V1
                                                                              6JW2 Andreas zweiter
                                                                       Teil.docD06_DRAFT_V15ejw



                                                                                                  Formatted: Bullets and Numbering
       5.2.35.2.4 L1 VPN
Basic Layer 1 services may be characterized in terms that include:
   -   Connectivity: Between a pair of CEs.
   -   Capacity: For example, the bit rate for a TDM service or the capacity of a
       wavelength service.
   -   Transparency: Optically, frame-, or bit rate transparent connection, depending on
       the context
   -   Availability: The percentage of time that the quality of the service meets the agreed
       criteria. To achieve the required level of availability for the customer connections
       the service provider's network may use restoration or protected resources.
   -   Performance: For example, the number of error-seconds per month.
Furthermore, the Layer 1 services may be categorized based on the combination of
connectivity features in the data plane and service control capability features (control
plane) available to the customer. A CE is associated with the service interface between a
customer site and the network, and the categorization can be seen in the context of this
service interface as follows. A single connection between a pair of CEs provides a static
service. This classic private line service is achieved through a permanent connection. A
dynamic service is either a switched connection service, or a customer-controlled soft
permanent connection service.
A similar perspective may be put on multiple connections among a set of CEs. These could
be a static service, where the private network service consists of a mesh of permanent
connections. The dynamic service in that case is a dynamic private network service
consisting of any combination of switched connection services and customer-controlled soft
permanent connection services.
For a static service, the network provider is responsible for the management of both the
network infrastructure and the end user connections. For dynamic services, the network
provider is only responsible for the configuration of the infrastructure, while the customer
establishes connections dynamically (on-demand). Private network on-demand services
can be enhanced so that multiple private networks are supported across the Layer 1
network as Vvirtual Pprivate Nnetworks. These are Layer 1 Virtual Private Networks (L1
VPNs). Compared to the static service, the L1VPN service has features such as a separate
policy per VPN, and distribution of information about which CEs can participate in which
VPNs.
The standardization of architecture and service requirements on the Llayer 1 VPN has
recently been reviewed. Within the ITU-T, Study Group 13 has been working on a draft
recommendation for generic requirements and architectures for L1 VPN. The proposed
definition for L1 VPN is: A network formed by a set of customer edges (CE) and network
resources. Later in the document, L1 VPN is referred to as “a VPN service whose whichs
U-plane (resource plane) operates at layer 1. The L1 VPN service connects a number of
CEs with point-to-point connections. These connections, either optical or TDM, could be
permanent, soft-permanent, or switched”.
Apart from standardization attempts, this issue has also been approached from the view of
an expected paradigm shift in optical networking due to technology breakthrough. Various



                                      Page 78 of 168
                             IST IP NOBEL "Next generation
                                                                              Title of the document
                             Optical network for Broadband
                                  European Leadership"                   e1686d81-2994-497c-93f6-
                                                                  b2e913b07f59.docD06_DRAFT_V1
                                                                              6JW2 Andreas zweiter
                                                                       Teil.docD06_DRAFT_V15ejw




untraditional services (“lightwave services” executing the data transport on a corresponding
physical layer VPN) may be exploited in the first L1 VPNs to appear: Bundled optical
channel, Long-reach hyperfine WDM channels, OCDM, Quantum encrypted channel, etc.
The connectivity of such “light wave services” between the VPN sites may be executed by
optical switches/cross connects or rearrangeable OADMs. Where resilience to failure is
critical to the VPN, this may be accomplished by fast optical switching. The L1 VPN can
meet high requirements on integrity and on availability performance, when combining some
of the light wave services with path diversity and optical performance monitoring.
Of course, traditional client formats and legacy equipment such as SDH may be used in an
L1 VPN according to the ITU-T definition. Services may be provisioned across Layer 1
networks using ASON/GMPLS protocols. As a possible complement, the “lightwave
services” of L1 VPN should be outlined below:
      TDM-channel                                                                                Formatted: Bullets and Numbering

CE may connect to a L1 VPN via specified TDM channels,
      • Wavelength channel                                                                       Formatted: Bullets and Numbering

CE may connect to a L1 VPN via specified wavelength channels, according to grid
standards. This implies little problems, and wavelength routing is enabled by e.g. shifting
wavelength by OEO transponders. However, transportation of other “light wave services”
on legacy WDM systems usually requires non-inherent optical performance monitoring.
What the provider needs to transport the customer’s data are mechanisms for the
management of their power, wavelength, and bandwidth.
      • Bundled optical channel                                                                  Formatted: Bullets and Numbering

The span of two or more adjacent Grid frequencies is combined into a single, "bundled"
optical channel (say 4x50 GHz = 200 GHz) given a carefully designed dispersion. The
expanded optical channel space may carry digital clients with higher bit rates than
standard-grid optical channels do. The additional dispersion compensation needed can be
employed on this portion of spectrum only. A precise compensation of chromatic dispersion
may be achieved throughout the channel width by the use of relatively inexpensive fibre
Bragg gratings and a circulator. PMD should also be considered. In this way filtering of
standard channels is not affected by the bundling.
      • Hyperfine WDM channels                                                                   Formatted: Bullets and Numbering

An attractive feature of hyperfine channels is their ability of long-reach hops without need
for dispersion compensation of fibre. Regeneration of these signals is not required. The
major requirements on system equipment are a high wavelength stability of the lasers and
filters used, and a sufficiently low value of OSNR (by high spectral power density) to be
preserved throughout the hop length. The service may be used to interconnecting distant
customers with a modest bandwidth demand, e.g. 1 Gbit/s per channel.
      • OCDMA                                                                                    Formatted: Bullets and Numbering

Optical Code Division Multiple Access is essentially a modulation technique based on
physical components as sequentially written fibre Bragg gratings, but OCDMA may
superimpose a VPN on a general, fixed network topology. Features such as connectivity
mimic that of a network of stationary cellular phones. Connections are established between
the uniquely defined terminal equipment. This service could be valuable for customers



                                      Page 79 of 168
                                IST IP NOBEL "Next generation
                                                                                  Title of the document
                                Optical network for Broadband
                                     European Leadership"                   e1686d81-2994-497c-93f6-
                                                                     b2e913b07f59.docD06_DRAFT_V1
                                                                                 6JW2 Andreas zweiter
                                                                          Teil.docD06_DRAFT_V15ejw




demanding enhanced physical network confidentiality, and also when assignment of
connections and channel capacity is handled directly by the customer.
         • Quantum encryption                                                                        Formatted: Bullets and Numbering

Low attenuating, un-amplified optical channels can be used at the network-edge for secure
short-range transmission of digital encryption keys. At the ingress of core network the keys
are forwarded by robust techniques (channel scrambling and diverse connections). After
receive, the key is used for dynamic encryption of information to be carried by traditional
means on e.g. IP layer. Due to physical limitations, the use of a quantum-encryption is
confined to short distances. On access links amenable to eavesdropping this is efficiently
prevented by quantum encryption. Markets will be classified government information,
intellectual property protection, and financial transaction security 2
         • Variable bit rate                                                                         Formatted: Bullets and Numbering

This "pay-per-bandwidth" connection service can be applied in networks having bit rate
transparent optical channels. Opto-electrical interfaces such as OEXC are allowed in the
nodes. Degradation of distant transparent connections may be overcome by reducing bit
rate.
         • Best effort                                                                               Formatted: Bullets and Numbering

Closely related to "Variable bit rate". Transmission quality of long distance connections is
usually degraded, and these and secondary-ranked C-band channels may be
commercialised as “Best effort”.
         • Light source                                                                              Formatted: Bullets and Numbering

A light source with specified power and filtering allowances are distributed to customer
equipment for signal modulation within specified limits. The applications of a "light source"
service span from variable bit rate channels to advanced phase modulation.
         • Coherent channel                                                                          Formatted: Bullets and Numbering

The customer is offered a precisely stabilized wavelength with high coherence length, in
order to enable phase-modulated transmission. (See also "Light source" above.)
         • Low-noise channel                                                                         Formatted: Bullets and Numbering

Some channels having low noise characteristic may be extracted using OPM. The use of
low-noise channels is applicable to e.g. multilevel- and phase coding.
The performance of, and requirements on optical services can be quite different from the
requirements on traditional client formats. A listing of most significant parameters illustrates
some differences. Optical performance monitoring is required on services that might
interfere with other channels, when sharing physical resources such as EDFA, filter,
switches, etc.




2
    http://www.magiqtech.com/products/index.php




                                         Page 80 of 168
                                  IST IP NOBEL "Next generation
                                                                                    Title of the document
                                  Optical network for Broadband
                                       European Leadership"                  e1686d81-2994-497c-93f6-
                                                                      b2e913b07f59.docD06_DRAFT_V1
                                                                                  6JW2 Andreas zweiter
                                                                           Teil.docD06_DRAFT_V15ejw




Lightwave service     Reconfigurable    Scaleable        Transport,   Bit rate,    Performance
                                                         km           Gbit/s       monitoring
                                                         unrepeated
TDM - Channel         High              High             >600 km 10   Multiple     ES; SES
                                                         Gbps         of 140
                                                                      Mbit/s
WDM channel:          High              HighMedium       >600 km 10   2.5,10, 40   Power, FEC
traditional digital                                      Gbps
client on lambda
WDM channel:          L-Switched or     Part of system   >600 km 10   Client,      Power,
lambda routed         oeo converted                      Gbps         grid         wavelength
                                                                      depend
Bundled OCH           Custom wide-      Part of system   Typical      >40
                      band filter rqd
Hyperfine             Custom narrow     1000 channels    >1000        0.5 -–1      Carrier
WDM                   band filter rqd                                              frequency
OCDMA                 Inherent          Max 8 channels   Typical      10, Chip     Chip power
                                                                      limited
Quantum               No                No               60 km        key          N.A.
encryption                                                            exchange
Variable bit rate     Yes               High             >= typical                High rates
                                                                                   only
Best effort           Yes               Part of system   Typical                   No
Light source                                             Typical                   Required
Coherent channel      No                                 Typical                   Required
Low-noise channel     N.A.              No               Limited                   Y (SNR)

Table 282828: VPN-L1 lightwave services and significant properties.

Upper row (grey) shows traditional digital services, for reference. Blank positions: either
unsignificant,Either unsignificant or no specific problem/requirements are expected.




                                          Page 81 of 168
IST IP NOBEL "Next generation
                                           Title of the document
Optical network for Broadband
     European Leadership"              e1686d81-2994-497c-93f6-
                                b2e913b07f59.docD06_DRAFT_V1
                                            6JW2 Andreas zweiter
                                     Teil.docD06_DRAFT_V15ejw




        Page 82 of 168
                                                           IST IP NOBEL "Next generation
                                                                                                                                                                                   Title of the document
                                                           Optical network for Broadband
                                                                European Leadership"                                                                         e1686d81-2994-497c-93f6-
                                                                                                                                                      b2e913b07f59.docD06_DRAFT_V1
                                                                                                                                                                  6JW2 Andreas zweiter
                                                                                                                                                           Teil.docD06_DRAFT_V15ejw




Network                                            VPN - L3                           VPN - L2                                                                 VPN - L1
Service        Public IP          Business IP
Protocol
Transparency
Framing
Format at User                                                                                                                                                 SDH; ODU; SDH; ODU; SDH; ODU; SDH; ODU;
Side           IP                 IP               IP              IP                 Ethernet           Ethernet           Ethernet           Ethernet        Lambda    Lambda    Lambda    Lambda

                                                   IP -
VPN Identifier                                     Address         IP - Address                                             Trail Trace Trail Trace                                          Trail Trace Trail Trace
(User Service                                      MPLS -          MPLS -       VLAN /     VLAN /     VLAN /     VLAN /     Identifier + Identifier +                                        Identifier + Identifier +
Identifier)        n.a.           n.a.             Label           Label        MPLS Label MPLS Label MPLS Label MPLS Label special OH special OH                                            special OH special OH
Type of                                                      on demand/                                                     on demand/ on demand/                                            on demand/       on demand/
Provisioning       on demand on demand             permanent scheduled permanent                         permanent          scheduled scheduled permanent                     permanent      scheduled        scheduled
                                                                                                                                                                                             seconds,         seconds,
Max Set Up                                                         minutes/                                                 seconds/           seconds/                                      minutes,         minutes,
Time               n.a.           n.a.             hours           hours              hours              hours              minutes            minutes         hours          hours          hours            hours
                                                                                                                                                                                             seconds,         seconds,
Max Tear Down                                                      minutes/                                                 seconds/           seconds/                                      minutes,         minutes,
Time          n.a.                n.a.             hours           hours              hours              hours              minutes            minutes         hours          hours          hours            hours
                                                   > Max Set Up    > Max Set Up       > Max Set Up       > Max Set Up       > Max Set Up       > Max Set Up    > Max Set Up   > Max Set Up   > Max Set Up     > Max Set Up
                                                   Time + Max      Time + Max         Time + Max         Time + Max         Time + Max         Time + Max      Time + Max     Time + Max     Time + Max       Time + Max
Min Service                                        Tear Down       Tear Down          Tear Down          Tear Down          Tear Down          Tear Down       Tear Down      Tear Down      Tear Down        Tear Down
Holding Time       n.a.           n.a.             Time            Time               Time               Time               Time               Time            Time           Time           Time             Time
Service
Request
Blocking                                                                            not                  not                                                  not             not
Probability        n.a.           n.a.                         0                  0 applicable           applicable                     1%                 1% applicable      applicable           5,00E-04      5,00E-04
Configuration
Dynamics           no             no               yes             yes                no                 no                 yes                yes             no             no             yes              yes
Connectivity (at   packet         packet           packet          packet
"Primary           based          based            based           based
Level")            (m)p2mp        (m)p2mp          (m)p2mp         (m)p2mp            (m)p2mp            (m)p2mp            (m)p2mp            (m)p2mp         p2p            p2p            p2p              p2p
Service
Availability       low            high             high            high               low                high               low                high            low            high           low              high




Network                                            up to           up to              up to              up to              up to              up to           up to          up to          up to            up to
Service Bitrate 2 Mbit/s          20 Mbit/s        10Gbit/s        10Gbit/s           10Gbit/s           10Gbit/s           10Gbit/s           10Gbit/s        10Gbit/s       10Gbit/s       10Gbit/s         10Gbit/s
Capacity/
Bitrate
Granularity                   0                0               0                  0                  0                  0                  0                  0 155 Mbit/s    155 Mbit/s     155 Mbit/s       155 Mbit/s




QoS: Max
Latency            400 msec       100 msec         100 msec        10 msec            100 msec           100 msec           100 msec           100 msec        10 msec        3 msec         10 msec          3 msec


QoS:
Acceptable
Data Loss                   1%           0,10%            0,10%           0,01%               0,01%              0,01%
                                                                                                       0,01% BER<1E-12            0,01%                                       BER<1E-12      BER<1E-12        BER<1E-12
                                                                                                             BER /                                                            BER /          BER /            BER /
Performance        Packet     Packet Loss Packet     Packet Loss                                             Errored                                                          Errored        Errored          Errored
Monitoring         Loss Ratio Ratio       Loss Ratio Ratio       Frame Loss Frame Loss Frame Loss Frame Loss Seconds                                                          Seconds        Seconds          Seconds
Network/
Service                                                                                                                                                        According to According to According to According to
Visibility      No                Partially        Partially       Partially          No                 No                 Partially          Partially       SLA          SLA          SLA          SLA
User                                                                                                                                                                                         According to     According to
Configurability No                No               Partially       Partially          No                 No                 Partially          Partially       No             No             SLA              SLA
                                                                                                                                                               Bandwidth,     Bandwidth,     Bandwidth,       Bandwidth,
                   hold time - hold time -         hold time - hold time -            hold time,         hold time,         hold time,         hold time,      Time,          Time,          Time,            Time,
Billing            data traffic data traffic       data traffic data traffic          QoS                QoS                QoS                QoS             Distance       Distance       Distance         Distance
                   Application Application         Application Application            Application        Application        Application Application            Application Application       Application Application
Security           dependent dependent             dependent dependent                dependent          dependent          dependent dependent                dependent dependent           dependent dependent




Table 292929: Network service table. The QoS Max latency boxes on black bottom
are hard to satisfy since a 3 milliseconds delay at the speed of light imply a distance
of around 300 km between points of exchange. The values appearing in these boxes
are related to the requirements of the mirrored storage applications of chapter 4.




                                                                                  Page 83 of 168
                             IST IP NOBEL "Next generation
                                                                               Title of the document
                             Optical network for Broadband
                                  European Leadership"                    e1686d81-2994-497c-93f6-
                                                                   b2e913b07f59.docD06_DRAFT_V1
                                                                               6JW2 Andreas zweiter
                                                                        Teil.docD06_DRAFT_V15ejw




 5.3    Network service table
In the network service table above, the five services and their classes are specified in
different aspects. There are of course many ways of producing these services. The table is
not intended to specify how the service is produced, but there is a connection between the
functionalities offered and how they are produced.
VPN could mean two different things:
1) A set of interfaces, between which one can establish connections.
2) One or more connections, between such interfaces. (Thus a single point-to-point
connection, such as a Leased Line is here classified as a special case of L1 VPN.)
Most of the parameters in the table (type of provisioning, setup time, etc) refer to the
individual connections, not to the set of interfaces. The VPN services are divided into low
and high service availability. The VPN services have two provisioning classes: permanent
and on demand, represented by the leased line concept and the switched line concept.
For connection-less services (Public IP, Business IP), the “network service bitrate”
parameter means the bitrate of the access link. For connection-oriented services (L3-VPN,
L2-VPN, L1-VPN), it means the bitrate of a single connection. But in principle, all L2 and L3
service definitions are bitrate-independent.
L1 connections thus include SDH channels, OTN channels, “format-transparent digital
channels”, and wavelength channels.


       5.3.1 Performance parameters
The vertical axis in the Network Service table gives the performance parameters of the
network services. Here are definitions and comments on these.

                      5.3.1.1 Protocol transparency
What protocols can be accepted in the different layers i.e. the level of transparency. On L2
and L3 protocols will be defined for framing of the payload of the higher layers. On L1 the
framing of higher layer shall be defined but as L1 partly is analogue other parameters than
normally are included in the protocols have to be specified. Examples of these parameters
are modulation formats, physical interfaces and wavelength range. All of them will influence
the transparency in the network. The fibre media, layer 0, will also have an impact on
transmission reach and capacity and need to be specified.

                      5.3.1.2 Standards - VPN identifier
The corresponding standards for the used VPN identifier in the layer are specified. The
user might have specific requirement on a particular data framing, format, speed etc.

                      5.3.1.3 Provisioning
Provisioning is here defined by the following parameters:
   -   The type of provisioning service: permanent set-up, scheduled or on demand/
       request. The provisioning "on request", require operator intervention. The "on



                                       Page 84 of 168
                              IST IP NOBEL "Next generation
                                                                                Title of the document
                              Optical network for Broadband
                                   European Leadership"                   e1686d81-2994-497c-93f6-
                                                                   b2e913b07f59.docD06_DRAFT_V1
                                                                               6JW2 Andreas zweiter
                                                                        Teil.docD06_DRAFT_V15ejw




       demand" is automatic or user-controlled, e.g. implemented with a web interface or
       with a signalling protocol.
   -   Max set-up time is the initial set-up time of a new connection. “On request”
       connections have longer set-up times than “on demand” connections and the set-up
       time will of course depend on the level of pre-installation and automation.
   -   Max teardown time is the time from a user initiated connection release to the time
       when the network releases all the involved resources. Normally the user doesn't
       care about this parameter.
   -   Min service holding time is the minimum time (duration) of the network service.
       Holding time relates to the lease time of the network service but it can differ from
       Min time. The contrary Max service holding time of the network service could be
       limited, which can be different from Max lease time.

                      5.3.1.4 Blocking probability
Service request blocking probability is the probability of not being able to establish the
requested connection due to missing resources (applicable only to on demand services).

                      5.3.1.5 Configuration dynamics
The connection parameters, which can be modified without a teardown, are here labelled
configuration dynamics. The parameters are changed via either paperpaper interface, web-
interface or signalling. Parameters that could be changed dynamically are e.g. bit-rate,
QoS, connectivity and resilience.

                      5.3.1.6 Connectivity
In L1 this is a connection orientated point-to-point connection that could be used for private
lines. Packet based connectivity will be used in L2 and L3. This is a (multi-)point-to-
multipoint connectivity usable for broadcast, multicast and LAN.

                      5.3.1.7 Service Availability
The resilience in a network depends on the implementation of protection, restoration and
rerouting. Min availability is the uptime of the network in percentage, e.g. an excellent
network has 99,999 % (“five 9’s”). Max recovery time is the time from a network service
interruption to service restoration by means of the implemented protection or restoration
mechanism (e.g. the 50ms in SDH protection). Max outage time is the longest consecutive
outage time, seconds. Outage time is only counted as outage if the interruption for any
reason is longer than the recovery process.

                      5.3.1.8 Bit-rate
Bit rate is a measure of the network’s capacity to transmit information. Other parameters
may be applicable, depending on the technology, e.g. burst size. Max network connection
bit-rate gives the momentary available bandwidth. In a packet-based network, this may be
different from the nominal connection bit-rate.




                                       Page 85 of 168
                             IST IP NOBEL "Next generation
                                                                              Title of the document
                             Optical network for Broadband
                                  European Leadership"                  e1686d81-2994-497c-93f6-
                                                                 b2e913b07f59.docD06_DRAFT_V1
                                                                             6JW2 Andreas zweiter
                                                                      Teil.docD06_DRAFT_V15ejw




During a network connection with dynamic capacity, other specifications are relevant.
Dynamic range gives the difference between min and max available bandwidth. Capacity
or bit rate granularity will give the minimum step of capacity change. Max increment/
decrement time is the time from user request to the increase/decrease of the connected
bit rate.

                     5.3.1.9 QoS
The QoS parameters specify the data integrity during an existing connection. The set of
parameters for QoS depends on the technology used and the data transport format. The
most relevant parameters, in a packet based transport system, are latency and packet loss.
These two parameters can vary a lot and depends on network architecture, equipment
used and data transport distance. For many applications, latency and data loss are very
important to give the proper performance to a customer service. Often there is a trade-off
between latency and packet delivery since larger buffers can reduce packet loss at the
expense of aan increased latency. Also at the L1 level we can see this trade-off; by
increasing the redundancy using a data line coder (FEC) we can reduce data errors, but
with a somewhat later data delivery.
Max latency expresses how much time it takes for a data item to get from one designated
point to another. It could be defined as a mean value but an absolute measure of the
largest delay of transferred data is adequate for most time sensitive applications such as
voice and video in real time.
Acceptable data loss varies over the network layers. For L2/L3 services, packet loss (in
%) is most relevant, while BER is more interesting for a connection oriented L1 service.
For a network service, the corresponding parameters from the standard that defines the
service should be used. For a user service, service classes may be used as a simplification
Performance monitoring is the measurement of one or more parameters to describe or
guarantee the quality of the network service.
Policing will control the maximum traffic flow into the network to enforce the SLS
parameters.

                     5.3.1.10 Customer insight
Network /service visibility indicates whether the customer has a view of his service and
resources in the network.
User configurability indicates whether the customer has management access to his
resources and authority to configure the services he is using.

                     5.3.1.11 Billing
Billing is not a main issue in this report. The table gives some parameters, which could be
suitable as a base for billing of the network service.




                                     Page 86 of 168
                             IST IP NOBEL "Next generation
                                                                                                                                                                                                                                                                                                  Title of the document
                             Optical network for Broadband
                                  European Leadership"                                                                                                                                                                        e1686d81-2994-497c-93f6-
                                                                                                                                                                                                                       b2e913b07f59.docD06_DRAFT_V1
                                                                                                                                                                                                                                   6JW2 Andreas zweiter
                                                                                                                                                                                                                            Teil.docD06_DRAFT_V15ejw




                      5.3.1.12 Security
Security includes many different things e.g. authentication, access control and data
security.


 5.4    Mapping network services groups to applications




                                                                                                                                                                                                                                                                                                              - Narrowband Voice, data (VoIP,...)




                                                                                                                                                                                                                                                                                                                                                                            - Digital distribution, digital cinema
                                                                                                                                                                                                                                                - Video Broadcast (IP-TV)
                                                                                         - Asyncrhonous Mirroring




                                                                                                                                                                                                                                                                                                                                                                                                                                          Tele-medicine/diagnostic
                                                                                                                    - Synchronous Mirroring
                                                                   - Storage on Demand
                                             - Back-Up / Restore




                                                                                                                                                                                                                            - Video on Demand




                                                                                                                                                                                                                                                                                                                                                                                                                     - Video conference
                                                                                                                                                                                                                                                                            - Video Download
                                                                                                                                              Grid computing
                                                                                                                                                               - Compute Grid




                                                                                                                                                                                                                                                                                               - Video Chat
                                                                                                                                                                                              - Utility Grid
                                                                                                                                                                                - Data Grid


                                                                                                                                                                                                               Multimedia




                                                                                                                                                                                                                                                                                                                                                    - Gambling
                                                                                                                                                                                                                                                                                                                                                                 - Gaming
                                   Storage




Public IP
Business IP
VPN - L3    permanent
            on-demand
VPN - L2    permanent, Hi avail
            permanent, Low avail
            on-demand, Hi avail
            on-demand, Low avail
VPN - L1    permanent, Hi avail
            permanent, Low avail
            on-demand, Hi avail
            on-demand, Low avail


       Table 303030: Network Service classes mapped to the different applications.
In the table, the light blue means that the application will run on this network service, the
dark blue gives a more efficient implementation, the white should be interpreted as this
service have no support for the application and the grey is just a separator between groups
of applications.
Below are given some comments on the mapping of applications vs. network services.


       5.4.1 Storage applications
Dedicated storage protocols (e.g. Fibre Channel, ESCON, FICON) have intrinsically the
majority of features typical of L2 protocols such as Ethernet, so L2-VPN is not a good
choice for Storage Applications. Back-up/restore applications do not need an excessive
availability and they have no short latency problem, so L1-VPN are suitable for this kind of
applications only where there is a need to transfer huge quantity of data. In other cases L3-
VPNs are more suitable for this kind of applications.




                                              Page 87 of 168
                              IST IP NOBEL "Next generation
                                                                                  Title of the document
                              Optical network for Broadband
                                   European Leadership"                     e1686d81-2994-497c-93f6-
                                                                     b2e913b07f59.docD06_DRAFT_V1
                                                                                 6JW2 Andreas zweiter
                                                                          Teil.docD06_DRAFT_V15ejw




For Storage on Demand (SoD) the situation is similar; the only difference is that for this
kind of application permanent network services have no sense for the variability and
unpredictability of the traffic.
For mirroring applications (synchronous and asynchronous), since they often require very
high QoS (e.g. short latency) and large bandwidth, L1-VPN should be suitable, even if, in
particular for asynchronous mirroring, L3-VPN might be a useful service.


       5.4.2 Grid
In general, for grid applications, permanent L1 connections are not suitable for carrying
such traffic, since a great number of nodes may require connections in a fully meshed
topology (the number of connection grows as the square with the numbers of nodes) and
so the cost of connectivity would become unsustainable. The grid service also needs a
very fine bandwidth-granularity, so an L1 service is not suitable, since we cannot reach a
bandwidth-granularity finer than VC12 (2 Mbit/s). Therefore VPNs at L2 and L3 are most
suitable to carry the traffic generated by compute, utility and in certain cases also data grid
services. For data grids, in general the requirement of small bandwidth granularity is not so
pressing as in the other two grid applications, so in case of large quantities of data to
transfer, L1-VPN may be more suitable.


       5.4.3 Multimedia
For all residential multimedia applications (Video on Demand, Video Chat, Video
Download), Public IP is a suitable network service. For some applications for which greater           Comment [POA4]: VoIP is not a multimedia
bandwidth is required (VoD, Video download) Business IP may be better.                                service, If VoIP is important as application, write
                                                                                                      separate section on “voice” where the complete
Video Broadcast and IP-TV represent a particular case. It is possible to imagine, that to             voice trend is described; fixed, mobile, VoIP,
                                                                                                      migration etc.
provide a Video broadcast or IP-TV service, we need a point of presence that “radiates” the
video to the customers. This point is usually a replication point located in the metro area
where the customers are. The link between the replication point and the customer site may
be provided by an L2-VPN or Public IP service. The video source is generated inside a
metro area, but the data transferred from the source of the video signal to the replication
points must be carried on a service with higher capacity and availability, such as an L1-
VPN to minimize excessive delays and too low availability.
Concerning telemedicine and tele-diagnostics; if it is on-line a few Mbit/s are necessary; as
well as high availability, short latency and low jitter, so L1 may be the best way to carry this
kind of traffic, in particular because economic issues are of less importance in comparison
to performance issues. Concerning off-line diagnostics, performance requirements are less
pressing, so in this case the choice Public or Business IP, will be good enough.




                                       Page 88 of 168
                               IST IP NOBEL "Next generation
                                                                                   Title of the document
                               Optical network for Broadband
                                    European Leadership"                      e1686d81-2994-497c-93f6-
                                                                       b2e913b07f59.docD06_DRAFT_V1
                                                                                   6JW2 Andreas zweiter
                                                                            Teil.docD06_DRAFT_V15ejw




 6        Traffic typology and relations concerning core and metro
          networks
Traffic typology deals with the study and analysis of the traffic based on traffic types or
traffic categories. The objective of this chapter is not only to identify the main traffic types
that are to be found flowing through core or metro networks, but also to infer the impact
that these different types of traffic have on the architectural design of the core and metro
networks.
Traffic classification and categorization can be made according different criteria. An
empirical approach could be based on the measurement, characterisation and modelling of
the traffic using metrics such as the application or the transport protocol. This kind of
approach will be followed in WP2 activities, and will be useful for validating and supporting
the preliminary traffic categorisation made in this document.
The traffic categorization approach used in this document has a more deductive
charactercharacteristic, as it will be mainly based on criteria that strongly depend on the
given services, as well as on their implementation and their social acceptance and
success. Each service will define its traffic generation phenomenon and the main
requirements for this traffic to be conveyed conveniently. Indeed, the actual implementation
of a given service with a concrete application or set of applications will relax or harden the
previous service requirements. E.g., different compression algorithms can be used for
encoding voice.
Different services implemented in several ways will have different social acceptance. For
instance, P2P paradigm has clearly imposed on files downloading applications. As a
consequence, traffic distribution along the networks is strongly affected by the P2P
paradigm.
Notwithstanding, this deductive approach, although theoretic, will also be based on
experimental evidence.


 6.1      Criteria for traffic classification
The following criteria may be used to make a traffic classification:
        Content orientation (depends on the service/application). These attributes are
         stackable, as they are preserved on traffic aggregates.
            o   Elasticity level refers to the level up to which the traffic original shape can
                be modified. Normally, communication services aspire to keep both the data
                and temporal integrity. In order to establish the elasticity level of a given
                service/application, it is useful to assess which of both integrities is more
                restrictive. Although different levels of elasticity can be defined, two opposite
                approaches are usually considered:




                                        Page 89 of 168
                                  IST IP NOBEL "Next generation
                                                                                            Title of the document
                                  Optical network for Broadband
                                       European Leadership"                          e1686d81-2994-497c-93f6-
                                                                              b2e913b07f59.docD06_DRAFT_V1
                                                                                          6JW2 Andreas zweiter
                                                                                   Teil.docD06_DRAFT_V15ejw




                         Inelastic traffic (a.k.a. stream traffic3) is generated by applications
                          whichapplications that have [Schulz] a timing relationship between
                          origin and destination, that is, temporal integrity overwhelms data
                          integrity. Therefore, the traffic generated by services addressed to
                          emulate virtual presence, where temporal integrity is the more
                          relevant consideration, is inelastic.
                         Elastic traffic is generated by applications where data integrity
                          overwhelms temporal integrity, therefore being rather tolerant to
                          delays and being able to adapt their data generation rate to network
                          conditions.
             o   Interactivity level. Even in one-way communications, usually a certain level
                 of interactivity exists, as the receiver can periodically feed-backfeedback to
                 the sender with signalling information. However, in this document,
                 interactivity will mean that communication is actively done in two-ways.
                 Although different levels of interactivity can be defined, two opposite
                 approaches can be considered:
                         Interactive applications are usually inelastic,inelastic; as they try to
                          simulate virtual presence by achieving an end-to-end delay lower
                          than a given threshold. This threshold depends mainly on the type of
                          interactivity: man-man, man-machine, machine-machine. However, if
                          this threshold is large enough, a certain grade of elasticity can be
                          tolerated.
                         Non interactiveNon-interactive applications do not require
                          continuous interaction amongst different ends of the communication
                          channel. Two examples of this kind of applications are e-mail and
                          video broadcasting.
             o   Traffic Asymmetry express the relationship between the traffic
                 volumevolumes sent and received by an application. For instance, web
                 browsing produces a highly asymmetric traffic, while audio or video
                 conference applications traffic is practically symmetric. Asymmetry is
                 strongly related to the interactivity level. Interactive applications usually
                 produce symmetric traffic, whereas non-interactive ones usually show a
                 more asymmetrical pattern. However, there are important exceptions to this
                 relation. For instance, P2P traffic, which has a low interactivity level, shows
                 a certaincertain symmetry on the traffic aggregate. When traffic is
                 asymmetric, it is interesting to distinguish which direction is more dominant.
                 The upstream direction is defined, from the user's perspective, from the user
                 location toward the remote destination. Conversely, downstream direction is
                 defined from the remote destination toward the user location.
             o   Service Availability can be defined as the percentage of time when service
                 is operating normally. Although communication necessities can inherently
                 impose different levels of service availability, normally, more stringent
                 constraints come exogenously from specific users. So, mission critical
3
 Although inelastic traffic is also referred as stream traffic [Robert2], this term will not be used in this
context as can be confused with streaming traffic, which is a particular type of inelastic traffic.




                                             Page 90 of 168
                                IST IP NOBEL "Next generation
                                                                                   Title of the document
                                Optical network for Broadband
                                     European Leadership"                    e1686d81-2994-497c-93f6-
                                                                      b2e913b07f59.docD06_DRAFT_V1
                                                                                  6JW2 Andreas zweiter
                                                                           Teil.docD06_DRAFT_V15ejw




                 services, which are normally related with business customers, are those
                 with the highest service availability requirements. Service availability
                 strongly depends, but not only, on network availability. It is possible to
                 achieve different grades of network availability by using different resilience
                 mechanisms. A first classification based on the service availability could be
                 made by distinguishing standard availability and high (or special) availability.
             o   Bandwidth necessities strongly depend on the given service and on its
                 resolution. The latter leads to a concrete virtual presence sensation, and
                 therefore, to a certain perceived QoS. The former leads to very different
                 necessities. For instance, video and audio streaming bandwidth differ on
                 several orders of magnitude. Usually distinction is made between
                 narrowband (low bandwidth) and broadband (high bandwidth). Although the
                 frontier between narrowband and broadband is not unanimously defined4,
                 the former can be associated, for instance to voice services or those with a
                 similar bandwidth, and the latter to services with a similar bandwidth that
                 video ones.
             o   Traffic variability (burstnessburstiness) depends not only on the service
                 nature, but also on the concrete implementation. A first classification could
                 be made by distinguishing between Constant Bit Rate (CBR) applications
                 and Variable Bit Rate (VBR) applications. Elastic applications are able, by
                 definition, to change the used bandwidth in function of the available network
                 resources. On the other side, inelastic applications were implemented, in the
                 old days, with CBR algorithms. However, the evolution of compression
                 algorithms, speciallyespecially in video applications, have allowed the
                 optimisation of bandwidth by suppressing redundant (both objective and
                 subjective) information, resulting in media applications not needing constant
                 bandwidth to work properly.
The importance of these criteria is relative, and depends on the considered point of view.
From the applications point of view, QoS is one of the factors to be considered. Thence,
elasticity and interactivity levels are, maybe, the parameters that have a more clear impact
on QoS. A classification of traffic generated by the different applications, considered in
previous chapters, based on the first above criteria could be the one depicted in Figure
18Figure 19Figure 19.




4
    In NOBEL, narrowband means less than 64Kbps.




                                         Page 91 of 168
                                 IST IP NOBEL "Next generation
                                                                                          Title of the document
                                 Optical network for Broadband
                                      European Leadership"                        e1686d81-2994-497c-93f6-
                                                                           b2e913b07f59.docD06_DRAFT_V1
                                                                                       6JW2 Andreas zweiter
                                                                                Teil.docD06_DRAFT_V15ejw




        Figure 181919: Classification of traffic generated by reference applications


        Topological and geographical distribution (these topics must be specially
         broached from a core/metro network point of view).
             o   Traffic generation dynamics express whether applications are expected to
                 continuously generate traffic or whether traffic is generated in certain
                 periods of time. For instance, media distribution applications generate
                 traffic 5 almost continuously; that is to say, traffic generation dynamics is
                 relatively low. On the contrary, telephony applications generate traffic only

5
  Multicast or broadcast applications are being referred here. If the potential destination group is
large enough, most of the time there will be at least a user receiving traffic.




                                           Page 92 of 168
                                    IST IP NOBEL "Next generation
                                                                                            Title of the document
                                    Optical network for Broadband
                                         European Leadership"                          e1686d81-2994-497c-93f6-
                                                                                b2e913b07f59.docD06_DRAFT_V1
                                                                                            6JW2 Andreas zweiter
                                                                                     Teil.docD06_DRAFT_V15ejw




                   when users are willing to communicate; namely, traffic generation dynamics
                   is relatively high. This traffic generation dynamics is related with the
                   interactivity level. If man-man interactivity is required, traffic will be
                   generated only when all participants are willing to communicate. Besides,
                   when aggregating traffic, this dynamics is blurred.
               o   Traffic sources scattering: Sources of traffic of a given application may be
                   very distributed or very located mainly depending on whether the
                   communication ends are at the same level (P2P paradigm) or whether there
                   is a dominion relation (client-server paradigm). On the one hand, P2P
                   paradigm rouses traffic sources are verymuch spread. On the other hand,
                   traffic generated by applications that follow the client-server paradigm is not
                   so distributed, as server functionality (that is supposed to generate more
                   traffic than the client) is used, normally, by a reduced set of users due to its
                   bigger complexity and speciality. The latter is, of course, very relative as
                   some well known applications have become so popular that it is very easy to
                   perform the server role. That would be the case, for instance, of web traffic,
                   that although following the client-server paradigm, server functionality is
                   very scattered through the Internet community. On the contrary, media
                   streaming distribution servers are, for instance, more located.
               Therefore, it would be possible to define, at leas, three different levels of traffic
               sources scattering:
                           Completely distributed sources (e.g. Telephony)
                           Relatively distributed sources (e.g. Web browsing)
                           Relatively located sources (e.g. Video on Demand)
               o   Local properties of traffic: It has been shown that there is indeed
                   dependence of the traffic demands on the geographical distance between
                   the origin and the destination of the demand. This dependence is quite clear
                   in voice traffic and although it doesn’t seem to be so in IP traffic, recent
                   studies have shown the existence of local properties in P2P traffic [Robert1],
                   [Chu], [Gummad], which may change the geographical traffic distribution:
                           Voice traffic between i and j = Kv * Pi * Pj / Dij2 [Lievens]
                        where Kv is a constant, Pi and Pj are the population of i and j respectively
                        and Dij is the geographical distance between i and j.
                           IP traffic between i and j = Ki * Hi * Hj [Lievens]
                        where Ki is a constant and Hi and Hj are the number of Internet hosts of i
                        and j respectively.
Any combination of the previous attributes can be present in the different types of traffic.
Due to this fact, instead of inspecting every type of traffic, the identification of the dominant
applications in a given traffic mix can be used to define the most relevant traffic types.
Hence, currently the more relevant traffic types would be (due to their differences, the
traffic has been split into two different groups – residential and institutions6):

6
    Enterprises, Administration, Public Affairs, Scientific Affairs will be included.




                                               Page 93 of 168
                               IST IP NOBEL "Next generation
                                                                                     Title of the document
                               Optical network for Broadband
                                    European Leadership"                      e1686d81-2994-497c-93f6-
                                                                       b2e913b07f59.docD06_DRAFT_V1
                                                                                   6JW2 Andreas zweiter
                                                                            Teil.docD06_DRAFT_V15ejw




           Residential:
                o   Media distribution (video, radio)
                Attributes: Inelastic, non interactivenon-interactive, downstream dominant,
                high availability, high/low bandwidth, medium burst-ness, low generation
                dynamics, relatively located sources.
                o   P2P, e-mail (background traffic)
                Attributes: Elastic, non interactivenon-interactive, symmetrical, standard
                availability, variable bandwidth, reduced burst-ness, low generation
                dynamics on the aggregates, completely distributed sources.
                o   Web-browsing
                Attributes: Elastic, interactive, downstream dominant, standard availability,
                low bandwidth, reduced burst-ness, low generation dynamics on the
                aggregates, relatively distributed sources.
                o   VoIP, interactive gaming
                Attributes: Inelastic, interactive, symmetrical, high availability, low
                bandwdthbandwidth, reduced burst-ness, low generation dynamics on the
                aggregates, completely distributed source7.
           Institutions:
                o   Intra-institution (VPN, leased lines, SAN)
                Attributes: Inelastic, interactive, symmetrical 8, high availability, high band-
                widthbandwidth, high burst-ness, low generation dynamics, relatively
                located sources.
                o   Inter-institution (Web-browsing, web-hosting, e-business)
                Attributes: Elastic, interactive, upstream dominant, high availability, high
                bandwidth, high burst-ness, high generation dynamics, relatively distributed
                sources.
                o   Institution-to-customers (web-hosting, online shopping)
                Attributes: Elastic, interactive, downstream dominant, high availability, low
                bandwidth, reduced burst-ness, high generation dynamics, relatively
                distributed sources




7
 In case of gaming, this depends on the concrete architectural solution that a given game has
adopted.
8
 SAN traffic for backup purposes will be, generally, asymmetrical, as synchronous/asynchronous
WRITE operations will be probably mostly directed towards a data center location. However, this
asymmetry may change during the data recovery (READ operations).




                                         Page 94 of 168
                               IST IP NOBEL "Next generation
                                                                                   Title of the document
                               Optical network for Broadband
                                    European Leadership"                    e1686d81-2994-497c-93f6-
                                                                     b2e913b07f59.docD06_DRAFT_V1
                                                                                 6JW2 Andreas zweiter
                                                                          Teil.docD06_DRAFT_V15ejw




    6.2    Traffic classes
Although there are many criteria to classify traffic, as has been indicated previously, it is
useful to do it in function of the QoS requirements, as different applications show
completely different sensitivity to the QoS.
A practical way to broach the QoS sensitivity analysis is to allocate a utility function
[Shenker] for each application in such a way that the utility function describes how the
performance of an application depends on the received Quality of Service.
The most relevant criteria for QoS purposes isare the inelastic/elastic classification as it is
the minimum classification to satisfy the QoS requirements of Internet traffic [Roberts2]. In
order to produce a more detailed classification, a further criterion can be introduced
(interactive and non-interactive), which produces the following four classes of traffic:


                                            Elastic                 Inelastic
             Interactive               Elastic interactive         Real Time
                                        (Transactional)         (Conversational)
             Non interactive         Elastic non interactive        Streaming

            Table 313131: Traffic Types
                                                                                                       Formatted: Bullets and Numbering
          5.2.1.6.2.1 Inelastic traffic
Inelastic traffic has a timing relationship between source (origin) and sink (destination);
that is, the sink must reproduce the timing relationship that existed at the source. The most
common examples of inelastic traffic are real time conversational traffic, audio and motion
video. Inelastic traffic can be interactive, this applications are called real time, where there
is a "tight" timing relationship between source and sink, or non interactivenon-interactive,
called streaming (playback), where the relationship is less strict.
                                                                                                       Formatted: Bullets and Numbering
                       5.2.1.16.2.1.1 Real time applications
    Real time applications try to simulate virtual presence because of the nature of the
     data they transmit. This way, these applications expect the data to arrive with a very
     low given delay and with the same rhythm (that is to say, with no delay variation) that
     they were generated so that practically no distance between the ends of the
     communication's channel be perceived by the application. Most audio and video
     distribution applications are implemented with buffers to be rather tolerant of delay
     variation requirement violations. Namely, these applications (e.g., telephony, gaming
     and leased line emulation) expect circuit-switched service.
     Consequently, real time applications need their data to arrive within a given delay
     bound: if packets arrive within this bound, real time applications perform in a good
     manner. However, packets that arrive out of the given bound are useless and produce
     performance degradation. This way, no difference between packet losses and delay
     violation is found at the application layer.
     There are two types of real time applications:



                                        Page 95 of 168
                                IST IP NOBEL "Next generation
                                                                                      Title of the document
                                Optical network for Broadband
                                     European Leadership"                      e1686d81-2994-497c-93f6-
                                                                        b2e913b07f59.docD06_DRAFT_V1
                                                                                    6JW2 Andreas zweiter
                                                                             Teil.docD06_DRAFT_V15ejw




       Hard real-time applications: While useful bandwidth9 perceived at the application
        layer is sufficient to meet the required delay bounds, the application performance is
        constant. But if the bandwidth becomes lower than the needed one in order to
        satisfy the required delay bounds, performance falls to zero.
        The utility curve of an application with hard real time requirements is represented in
        the Figure 19Figure 20Figure 20 as a step function:

                       Utility       U(w)




                                                                          w
                                                           Bandwidth
Figure 192020: Utility of a “hard” real time application as a function of the bandwidth
       Adaptive real-time applications: These applications are able to partially adapt to
        the network conditions. In case of network congestion, these applications are able
        to adjust their transmission rate by coding real-time data with a higher compression
        rate but a minimum performance degradation or by marking the least significance
        packets with a higher discard priority so that network can, when possible, forward
        these packets (and provide a higher performance) or discard them when congestion
        is taking place.
        Adaptive real-time applications have an intrinsic bandwidth requirement in order to
        work in an appropriate manner. With higher bandwidths, the performance of these
        applications is improved. Nevertheless, the signal quality that humans need is very
        close to the intrinsic bandwidth requirement. So, the incremental utility of additional
        bandwidth is slight. Also, with very small bandwidths, the signal quality is
        unbearably low.
        The utility curve of an application with adaptive real-time requirements can be
        depicted as a smoothed step function of the bandwidth, as the one of the Figure
        20Figure 21Figure 21.




9
 From now on, bandwidth requirements will be considered at the application layer; that is, useful
bandwidth perceived by the application.




                                          Page 96 of 168
                              IST IP NOBEL "Next generation
                                                                                 Title of the document
                              Optical network for Broadband
                                   European Leadership"                    e1686d81-2994-497c-93f6-
                                                                    b2e913b07f59.docD06_DRAFT_V1
                                                                                6JW2 Andreas zweiter
                                                                         Teil.docD06_DRAFT_V15ejw




                         Utility    U(w)




                                                                       w
                                                          Bandwidth
    Figure 202121: Utility of an adaptive real-time application as a function of the
                                      bandwidth
                                                                                                     Formatted: Bullets and Numbering
                       5.2.1.26.2.1.2 Streaming applications
Streaming applications have less stringent requirements than real time ones, as they are
not interactive. These applications are more tolerant to delay than real time applications.
Therefore, buffers used for avoiding delay variation can be much longer. Nonetheless, data
generation is still independent of the network state. So, these applications still have an
intrinsic bandwidth requirement. Therefore, performance degrades when the useful
bandwidth perceived falls below the intrinsic bandwidth during time enough to empty the
buffer.

                                                             Cumulative Useful
                                                                Bandwidth
   Buffer occupancy (t)
                                                                Cumulative
                                                             Reproduction rate

         Play                            Pause
                 t                                                    Time
                     Figure 212222: Working of a streaming application
                                                                                                     Formatted: Bullets and Numbering
       5.2.2.6.2.2 Elastic traffic
Elastic traffic is rather tolerant to delays, as the applications that generate elastic traffic
can adapt their data generation rate to network conditions. Traditional data applications like
file transfer (FTP),; e-mail and remote terminal (TELNET) are examples of elastic
applications.




                                       Page 97 of 168
                              IST IP NOBEL "Next generation
                                                                                 Title of the document
                              Optical network for Broadband
                                   European Leadership"                    e1686d81-2994-497c-93f6-
                                                                    b2e913b07f59.docD06_DRAFT_V1
                                                                                6JW2 Andreas zweiter
                                                                         Teil.docD06_DRAFT_V15ejw




Although some applications like e-mail or FTP can tolerate very high delays, elastic
applications normally require a limited total transaction time, which varies from one
application to another and highly depends on the user patience (e.g., remote terminal or
web browsing require low total transaction times in order to avoid users lose their temper).
The former are elastic non-interactive applications, while the latter are called transactional
interactive applications.
Elastic applications require a reliable traffic transmission, without losses at the application
level. Because of this, a protocol like TCP is needed since it is able to retransmit lost
packets. TCP is also able to adapt the transfer rate to the available bandwidth, allowing the
application to work with whichever bandwidth even when the network is congested.
However, guaranteeing a minimum bandwidth to this traffic enables TCP applications to
work in a better manner.
So, a high available bandwidth improves the performance of elastic applications. However,
there is a decreasing incremental improvement due to incremental increases in bandwidth;
that is known as saturation, and is due to the fact that getting a total transaction time lower
than a given virtual presence threshold is not useful as human (or machine) perception has
its own limits. Considering that perceived useful bandwidth and total transaction time are
inversely proportional, the utility curve of an application with elastic requirements and no
interactivity constraints looks like the one in Figure 22Figure 23Figure 23.

                        Utility     U(w)



                                                      Virtual presence
                                                         threshold w

                                                          Bandwidth
  Figure 222323: Utility of an elastic non-interactive application as a function of the
                                        bandwidth
On its turn, an elastic application with interactivity requirements have to obtain a minimum
bandwidth in order to avoid




                                       Page 98 of 168
                               IST IP NOBEL "Next generation
                                                                                    Title of the document
                               Optical network for Broadband
                                    European Leadership"                     e1686d81-2994-497c-93f6-
                                                                      b2e913b07f59.docD06_DRAFT_V1
                                                                                  6JW2 Andreas zweiter
                                                                           Teil.docD06_DRAFT_V15ejw




                     Utility       U(w)



                                      Patience          Virtual presence
                                     threshold             threshold w

                                                         Bandwidth
    Figure 232424: Utility of an elastic interactive application as a function of the
                                       bandwidth


 6.3     Traffic characteristics per application and per user
The different characteristics of traffic (diary profile, evolution, growth of main traffic classes
and specific reference applications) should be studied, making a characterization of both
the applications and the users, in order to be able to make traffic forecasts, network
provisioning and define the QoS classes needed in a given network, as the characteristics
of traffic impact on the core and metro networks architecture.
The traffic measurements and characterization work under development in WP2 will
provide experimental evidence of the main traffic types in core and metro networks. As
output, WP2 will provide traffic profiles by application, as the one showed as an example in
this chapter.


       6.3.1 Metro Networks
All figures of this section have been generated by using measurements performed by the
SMARTxAC project (http://www.ccaba.upc.es/contingut.php?dir=Projects/SMARTxAC)
under a collaboration agreement between the Advanced Broadband Communications
Centre (CCABA) of the Technical University of Catalonia (UPC) and the Supercomputing
Centre of Catalonia (CESCA) over the Catalan R&D Network, Anella Científica,
(http://www.cesca.es/en/comunicacions/anella.html). This network can be considered as a
metro network due to its wide range and the presence of multiple institutions. Daily profiles
belong to March 23rd of 2004. Weekly profiles belong from 22nd to 28th of same month.
The following figures show a diary profile by application of a metro network. Both figures
demonstrate that the main traffic types, both incoming and outgoing, are web traffic and
P2P traffic, although incoming and outgoing traffic show different characteristics. Moreover
in Figure 24Figure 25Figure 25 and Figure 25Figure 26Figure 26, it can be observed that
unknown traffic and P2P traffic have a very similar profile (they parallel evolve – red and
green lines), meaning that most unknown traffic is indeed P2P traffic disguised by using
random or well-known (e.g. port 80) ports.




                                        Page 99 of 168
                           IST IP NOBEL "Next generation
                                                                             Title of the document
                           Optical network for Broadband
                                European Leadership"                 e1686d81-2994-497c-93f6-
                                                              b2e913b07f59.docD06_DRAFT_V1
                                                                          6JW2 Andreas zweiter
                                                                   Teil.docD06_DRAFT_V15ejw




     Figure 242525: Example of diary traffic profile by application (cumulative)




            Figure 252626: Example of diary traffic profile by application


The following figures (Figure 26Figure 27Figure 27) of the weekly traffic of the same
network show that week days and weekend days have a completely different profile that
should be separately studied and taken into consideration when making traffic
characterization.




                                   Page 100 of 168
                              IST IP NOBEL "Next generation
                                                                                Title of the document
                              Optical network for Broadband
                                   European Leadership"                   e1686d81-2994-497c-93f6-
                                                                   b2e913b07f59.docD06_DRAFT_V1
                                                                               6JW2 Andreas zweiter
                                                                        Teil.docD06_DRAFT_V15ejw




           Figure 262727: Example of a weekly traffic profile by application


Another significant characteristic of the different types of traffic is the asymmetry (for the
following figures, the asymmetry has been calculated by dividing the measured outgoing
traffic by the incoming one for every interval). It can be observed that for a given network,
the different types of traffic present different asymmetries.
On the left of Figure 27Figure 28Figure 28, web traffic asymmetry is showed. From this
figure, two different behaviours can be easily appreciated. On labour period (from 8:00 to
20:00), this network acts as content consumer, as downstream direction is dominant
(asymmetry is lower than 1). There must be emphasised that asymmetry is, on this labour
period, very uniform. On rest period (rest of time), network acts as content provider, as
upstream direction is dominant (asymmetry is bigger than 1).
On the right of same figure, P2P well-known applications asymmetry is showed. From the
figure can be deduced that this network acts normally as content provider.




                Figure 272828: Traffic asymmetry of web and P2P traffic




                                      Page 101 of 168
                              IST IP NOBEL "Next generation
                                                                                 Title of the document
                              Optical network for Broadband
                                   European Leadership"                    e1686d81-2994-497c-93f6-
                                                                    b2e913b07f59.docD06_DRAFT_V1
                                                                                6JW2 Andreas zweiter
                                                                         Teil.docD06_DRAFT_V15ejw




               Figure 282929: Traffic asymmetry of FTP and e-mail traffic
On the left of Figure 28Figure 29Figure 29, FTP asymmetry is depicted. Once more, two
different periods can be easily distinguished. Although it can be deduced that this network
acts, regarding to FTP applications, as content provider, it is clear that this role is
emphasised on rest hours.
On the right of same figure, e-Mail symmetry is clearly showshowed.
The following Figure 29Figure 30Figure 30 shows that different applications have different
packet sizes, which may help to identify them and also to identify disguised P2P traffic. It is
clear that most of traffic classified as unknown looks like disguised P2P traffic. There
seems to be, notwithstanding, a clear difference in nocturne hours (from 20:00 to 0:00),
which may correspond to the existence of an authentic and legitimate application
performing its labours at these hours.




                     Figure 293030: Mean packet size by application




                                      Page 102 of 168
                                                                  IST IP NOBEL "Next generation
                                                                                                                                                              Title of the document
                                                                  Optical network for Broadband
                                                                       European Leadership"                                                     e1686d81-2994-497c-93f6-
                                                                                                                                         b2e913b07f59.docD06_DRAFT_V1
                                                                                                                                                     6JW2 Andreas zweiter
                                                                                                                                              Teil.docD06_DRAFT_V15ejw




                     6.3.2 Core Networks
The level of traffic aggregation in core networks produces different traffic profiles
whichprofiles that have to be taken into consideration. For example, Figure 30Figure
31Figure 31reflects a one weekone-week traffic profile of P2P applications based on the
measurements done on a transit router 10 in February 2003. In this case, traffic is very
stable; that is, low dynamics on the P2P traffic aggregates can be, for instance, inferred.
Further work on this topic must be addressed, whether suitable measurements are
available, in WP2.
                                           Traffic by P2P application                                                            P2P Traffic by application




                                                                                                 Joltid
                                                                                                 M2P2
                                                                                                 Freenet                                                Edonkey2000
     Traffic




                                                                                                 BitTorrent
                                                                                                 SoulSeek
                                                                                                                                                            80%
                                                                                                 Gnutella
                                                                                                 DirectConnect
                                                                                                 FastTrack
                                                                                                 Winmx
                                                                                                 Edonkey2000


                                                                                                                 BitTorrent                                           Winmx
                                                                                                                     4%                                                3%
                                                                                                                              SoulSeek           FastTrack
                                                   Sunday
                       Friday




                                Saturday




                                                               Monday
          Thursday




                                                                        Tuesday




                                                                                     Wednesday




                                                                                                                                4%                  9%


                                               Day of the week



                     Figure 303131: Example of weekly P2P traffic profile in a core network


                     6.3.3 User profile
WP2 will also provide users profiles for different types of users. As an example, Figure
31Figure 32Figure 32 shows the diary profile11 during the labourworking days (Monday to
Friday) for a typical residential broadband user. Aggregated upstream traffic is very uniform
during the whole labourworking day, whereas aggregated downstream traffic doubles up
during labour hours. A potential explanation could be based on that during all the day there
are P2P applications working (that is evident looking at the previous figure), and during
labour hours web traffic takes more relevance.




10
     No more details are provided as data are confidential.
11
     Download and upload traffic are not stacked but overlaid.




                                                                                  Page 103 of 168
                                             IST IP NOBEL "Next generation
                                                                                                                             Title of the document
                                             Optical network for Broadband
                                                  European Leadership"                                              e1686d81-2994-497c-93f6-
                                                                                                             b2e913b07f59.docD06_DRAFT_V1
                                                                                                                         6JW2 Andreas zweiter
                                                                                                                  Teil.docD06_DRAFT_V15ejw




                          Average broadband user profile
                                                                         2002
                     Traffic




                                                                                     Download (nw-us)

                                                                                     Upload (us-nw)
                                                                                             16:00




                                                                                                                     22:00
                                                                     10:00

                                                                             12:00

                                                                                     14:00




                                                                                                     18:00

                                                                                                             20:00
                                      2:00




                                                              8:00
                               0:00




                                              4:00

                                                     6:00




                                                                     Time

                    Figure 313232: Example of diary user traffic profile
An important conclusion is that aggregated upstream residential traffic is reaching the
levels of aggregated downstream residential traffic. That is, traffic is becoming more and
more symmetric. This however, has more impact on access and metro networks, as many
are designed for dominant downstream traffic. Core networks, on the contrary, work with so
aggregated traffic that traffic asymmetry blurs.
This conclusion is supported also by Figure 32Figure 33Figure 33, which shows the
evolution of traffic asymmetry12 of the average residential dial-up user (narrowband user)
during the average labour day of September. It can be seen that during 2001 asymmetry
was rather uniform. In 2002, this asymmetry kept going during labour hours. However,
during nocturnal hours, this asymmetry came down as a consequence of P2P files
exchange success. In 2003, this asymmetry descent is even more pronounced.




12
  This time, and on the contrary that in the previous section, asymmetry has been defined as the
quotient between the incoming traffic and outgoing traffic, as residential users consume more
content than they provide.




                                                            Page 104 of 168
                                                                                                IST IP NOBEL "Next generation
                                                                                                                                                                                                                                                  Title of the document
                                                                                                Optical network for Broadband
                                                                                                     European Leadership"                                                                                                               e1686d81-2994-497c-93f6-
                                                                                                                                                                                                                                 b2e913b07f59.docD06_DRAFT_V1
                                                                                                                                                                                                                                             6JW2 Andreas zweiter
                                                                                                                                                                                                                                      Teil.docD06_DRAFT_V15ejw




                                                                                                                                         Labour day

                                                   5,0

                                                   4,5
             Incoming traffic / Outgoing traffic




                                                   4,0

                                                   3,5

                                                   3,0
                                                                                                                                                                                                                                               2001
                                                   2,5                                                                                                                                                                                         2002
                                                   2,0
                                                                                                                                                                                                                                               2003

                                                   1,5

                                                   1,0

                                                   0,5

                                                   0,0
                                                         7:00
                                                                8:00
                                                                       9:00




                                                                                                                                                                                              0:00
                                                                                                                                                                                                     1:00
                                                                                                                                                                                                            2:00
                                                                                                                                                                                                                   3:00
                                                                                                                                                                                                                          4:00
                                                                                                                                                                                                                                 5:00
                                                                                                                                                                                                                                        6:00
                                                                              10:00
                                                                                      11:00
                                                                                              12:00
                                                                                                      13:00
                                                                                                              14:00
                                                                                                                      15:00
                                                                                                                              16:00
                                                                                                                                      17:00
                                                                                                                                              18:00
                                                                                                                                                      19:00
                                                                                                                                                              20:00
                                                                                                                                                                      21:00
                                                                                                                                                                              22:00
                                                                                                                                                                                      23:00




                                                                                                                                               Time


        Figure 323333: Evolution of traffic asymmetry of average dial-up user


 6.4    Impact of traffic on the network architecture

       6.4.1 Network switching paradigms
The two different network switchingnetwork-switching paradigms for transmitting messages
through a telecommunication network are circuit switching and packet switching. In circuit
switching networks, a dedicated channel or circuit is specifically established for the
duration of a given transmission. The resources remain dedicated to the circuit during the
entire transfer and the whole message follows the same path. In packet-switched networks,
the message is broken into packets. These packets, that can be differently routed, are
recompiled into the original message when reach the destination.
Network switching paradigm obviously affects to the network realization and the final
network architecture, and therefore must be considered as a very important implementation
decision.
Circuit switching networks are aimed for communications that require data to be
transmitted in real-time, and preserves rather well the traffic shape, being therefore more
suitable for conveying inelastic traffic. Packet switching networks are more efficient as
resources are not solely allocated to a given communication but shared amongst several
ones. The drawback is that the traffic shape is modified, following that a packet switching
network is more appropriate for elastic traffic.
Circuit-switching networks are, by definition, connection-oriented networks, whereas packet
switchingpacket-switching network are essentially connectionless (a.k.a. datagram
switching) unless a higher-level protocol be used for providing a virtual circuit through the
packet network.




                                                                                                                              Page 105 of 168
                              IST IP NOBEL "Next generation
                                                                                Title of the document
                              Optical network for Broadband
                                   European Leadership"                   e1686d81-2994-497c-93f6-
                                                                   b2e913b07f59.docD06_DRAFT_V1
                                                                               6JW2 Andreas zweiter
                                                                        Teil.docD06_DRAFT_V15ejw




Optical networks, that are the target of NOBEL project, are nowadays implemented as
circuit switching networks. However, optical networks are expected to evolve from adopting
the circuit switchingcircuit-switching paradigm to the adoption of packet switching (OPS) or
even a mixture of both (OBS).
This means that, at least for some years, optical networks will continue being circuit
oriented and therefore more suitable for conveying inelastic traffic, although of course,
perfectly able to transport elastic traffic. As core networks will normally be optical based,
circuit switchingcircuit-switching paradigm will prevail in these networks. However, in metro
networks a higher range of technical possibilities can be deployed. Therefore, metro
networks can more easily be packet switching based.


       6.4.2 Topology of the network
When designing the network, a given topology must be chosen. Networks can be, for
instance, fully meshed, partially meshed, tree oriented, star oriented, or ring oriented.
Till now, ring topologies have imposed on core and metro networks due to the restoration
facilities that these topologies are able to provide.
As with connection dynamics, should a metro or core network be mostly devoted to media
distribution (in form of broadcast or multicast), a star or tree oriented topology could be
possibly the best solution, as it would minimise the amount of deployed resources.
However and due in part to the preponderance that P2P traffic is getting, and in part to be
able to satisfy high and variable traffic demands with an optimised solution, is expected
that current metro and core networks achieve a partially (if no fully) meshed topology.
There is also to take into account that the traffic mix in core and metro networks can be
different. For instance, the greatest part of telephony traffic will remain in metro networks,
due to the locality properties explained in previous section 2.3.22.3.22.3.2. On the contrary,
web traffic flows more globally. Besides, broadcast traffic, for instance, will constitute a
minimal part of the traffic mix both in metro and core networks. Figure 33Figure 34Figure
34 tries to illustrate this, by showing where these different traffic types come from.




                                      Page 106 of 168
                             IST IP NOBEL "Next generation
                                                                              Title of the document
                             Optical network for Broadband
                                  European Leadership"                   e1686d81-2994-497c-93f6-
                                                                  b2e913b07f59.docD06_DRAFT_V1
                                                                              6JW2 Andreas zweiter
                                                                       Teil.docD06_DRAFT_V15ejw




       Figure 333434: Distribution of traffic amongst core and metro networks


       6.4.3 Connection dynamics
In a connection oriented scenario, establishment of the connections can be done
permanently (or semi-permanently) or as they are needed (switched connections).
Permanent connections are used when traffic is regular and well known between a set of
ends or because connections cannot be automatically established (this is normally the
case of ATM and optical networks, for instance). On the contrary, switched connections are
wished to satisfy variable communication demands and to obtain a more optimal usage of
the network resources (by means of automated traffic engineering mechanisms). For being
able to dynamically establish connections, a complex control plane (CP) and network
management (NM) facilities are required. The CP/NM requirements necessary for being
able to provide dynamic establishment of connections are being studied in WP4.
Connection dynamics is very related with the traffic generation dynamics characteristic that
was addressed in the first section of this chapter.
Should a core or metro network be mainly devoted to media distribution (in form of
multicast or broadcast), this network could be easily optimised with no automatic
establishment mechanisms (that is to say, manually). Same can be applied when traffic
aggregates that flow in the network are well characterised and are rather constant. That
can be the case, for instance, of telephony long distance and international trunks.
Notwithstanding, core networks are expected to be more dynamic than current ones in
order to be able to provide a set of complex and different services on demand, being able



                                     Page 107 of 168
                            IST IP NOBEL "Next generation
                                                                             Title of the document
                            Optical network for Broadband
                                 European Leadership"                  e1686d81-2994-497c-93f6-
                                                                b2e913b07f59.docD06_DRAFT_V1
                                                                            6JW2 Andreas zweiter
                                                                     Teil.docD06_DRAFT_V15ejw




not only to serve to big customers (ISPs, ASPs, RSPs, big enterprises) but also even to
medium customers (SOHOs)
Besides, metro networks are expected to be much more dynamic than core networks, as
on demand services will be provided with less granularity than in core networks.
This connection dynamics will also increase when delivering special services as on
demand VPN connectivity. Figure 34Figure 35Figure 35 tries to illustrate this, showing how
a VIP customer (usually a big enterprise) may be based on different locations, connecting
directly to metro or even to core networks. Dashed lines are used to mean this high
dynamics.




 Figure 343535: Distribution of enterprise traffic amongst core and metro networks




                                    Page 108 of 168
                            IST IP NOBEL "Next generation
                                                                             Title of the document
                            Optical network for Broadband
                                 European Leadership"                  e1686d81-2994-497c-93f6-
                                                                b2e913b07f59.docD06_DRAFT_V1
                                                                            6JW2 Andreas zweiter
                                                                     Teil.docD06_DRAFT_V15ejw




 7      Emerging requirements for metro and core networks

 7.1    Introduction
The following sections describe a selected list of high-level requirements for next-
generation metro and core networks, as derived in NOBEL. Special attention is brought on
issues that relate to the Next Generation Network (NGN) project supported by ITU-T, in
order to facilitate the international coordination of research and development in the near
future on next-generation networks and services.




                                    Page 109 of 168
                              IST IP NOBEL "Next generation
                                                                              Title of the document
                              Optical network for Broadband
                                   European Leadership"                  e1686d81-2994-497c-93f6-
                                                                  b2e913b07f59.docD06_DRAFT_V1
                                                                              6JW2 Andreas zweiter
                                                                       Teil.docD06_DRAFT_V15ejw




   Figure 353636: Detailed requirement schematic in NOBEL based on the global
                  clustering proposed by the NGN project at ITU
With requirement list being shared across the NOBEL project, a longer requirement list is
formed using clustering proposed by ITU NGN. Figure 35Figure 36Figure 36 shows only
the complexity in requirement structuring, for which an overall arbitration is being sought.
Therefore, the following sections regroup NOBEL requirements into three main categories
as suggested by ITU NGN, Figure 35Figure 36Figure 36:
      Architecture
      Network requirements
      Service requirements




                                     Page 110 of 168
                             IST IP NOBEL "Next generation
                                                                             Title of the document
                             Optical network for Broadband
                                  European Leadership"                  e1686d81-2994-497c-93f6-
                                                                 b2e913b07f59.docD06_DRAFT_V1
                                                                             6JW2 Andreas zweiter
                                                                      Teil.docD06_DRAFT_V15ejw




These categories bring forward technological, architectural, and administrative challenges.
Furthermore, the explicitly numbered NOBEL requirements are embedded in subsections
that are also headlined as proposed by ITU NGN.


 7.2    Architecture

       7.2.1 Functionally separate architecture
The overall architecture shall functionally separate the following functional planes, as
shown in Figure 36Figure 37Figure 37:
      Transport plane
      Control plane
      Management plane
      Application plane




                                     Page 111 of 168
                             IST IP NOBEL "Next generation
                                                                               Title of the document
                             Optical network for Broadband
                                  European Leadership"                    e1686d81-2994-497c-93f6-
                                                                   b2e913b07f59.docD06_DRAFT_V1
                                                                               6JW2 Andreas zweiter
                                                                        Teil.docD06_DRAFT_V15ejw




     Figure 363737: Transport, control, and management planes in the network.
  As shown in the diagram, a business/service plane can be foreseen on the top
  of classical Management-Control-Transport planes. The user network can have a web-
  interface with the business/service plane. With this interface the user can send the
  Service Level Specifications (SLSs) like availability, QoS parameters, etc. to the
  Network operator. In the figure the business/service plane is separated from
  Management plane just for the sake of clarity in functionality. This Plane can be
  integrated in the Management plane and can be used as additional interface to the user
  network from management plane.



       7.2.2 Applications independent architecture
The so-called end-to-end principle states that the network must not be constrained by any
application.


       7.2.3 IP infrastructure and migration issues
The Internet transport infrastructure is moving towards a model of high-speed routers
interconnected by intelligent optical networks. A consensus is emerging in the industry on
utilizing an IP-centric control plane within optical networks to support dynamic provisioning
and restoration in the overall infrastructure.
Migrations issues are related essentially to backward compatibility to existing network
infrastructures. In order to facilitate seamless network migration, a promising direction for
network evolution lies in the migration of the switching technologies into the optical domain
in order to exploit the properties of optical technology. It allows supporting progressively
the optical transmission capacity. It may also cover the migration towards transparent




                                      Page 112 of 168
                              IST IP NOBEL "Next generation
                                                                                Title of the document
                              Optical network for Broadband
                                   European Leadership"                   e1686d81-2994-497c-93f6-
                                                                   b2e913b07f59.docD06_DRAFT_V1
                                                                               6JW2 Andreas zweiter
                                                                        Teil.docD06_DRAFT_V15ejw




switching and optical packet/burst switching paradigms, i.e. migration of circuit-based
services.
In particular, the Enterprise area seems to drift towards an enormous diversity of networks
and protocols, from support of legacy leased lines to emerging Grid and storage
applications. Multi-protocol capable LANs of enterprise quality from layer 1 (optical) to IP
appear to be necessary.


 7.3     Network requirements
                                                                                                    Formatted: Bullets and Numbering
       7.3.1 Data plane requirements

       7.3.27.3.1
- Meet BER requirements
One of the best ways to measure the data-transmission quality is to determine the bit-error-
rate (BER) characteristics. BER is greatly influenced by the optical signal-to-noise ratio
(OSNR) of the system. OSNR is the ratio between the received signal and the additive
noise of the optical link. The higher the OSNR, the lower the BER.
To reduce network cost and complexity, carriers are adopting high-bandwidth optical
switching systems, which offer a scalable solution to control optical bandwidth, for instance,
in the Internet area. While specific requirements for optical switching equipments may differ
from carrier to carrier, the performance verification to monitor BER for service level
verification remains one of the key issues both in opaque and transparent optical networks.


- Adjustable up & down stream bandwidths
Adjustable bandwidth gives costumers the ability to request additional bandwidth at specific
times. For operators, it means the requirement to the ability to provision services quickly
and have them adjustable on the fly.
- Optical Transparency
In optical networks, transparency means first no OEO conversion of the optical signal.
Metro and core networks can benefit from lower costs and increased speed by reducing
OEO conversions and avoiding the high cost of deploying optical termination units. It is
also called photonic networking, which brings on new challenges such as the issue of
length in between optical amplifiers, while the number of O/E/O conversions is minimized.
To secure improved efficiency, network operators have to choices generally between O-E-
O equipment and all-optical, i.e. photonic, O-O-O equipment. However, new architectures
encompass both. With the improved efficiency of the new optical/photonic equipments, one
of the most important keys for network operators is to manage the high DWDM capacity
coupled with the transparent data transfer.


       7.3.37.3.2 Routing and Resilience
The network architecture should support differentiated routing and resilience features.



                                      Page 113 of 168
                               IST IP NOBEL "Next generation
                                                                                     Title of the document
                               Optical network for Broadband
                                    European Leadership"                       e1686d81-2994-497c-93f6-
                                                                        b2e913b07f59.docD06_DRAFT_V1
                                                                                    6JW2 Andreas zweiter
                                                                             Teil.docD06_DRAFT_V15ejw




       7.3.47.3.3 QoS Requirements

- Support of different Quality of Service
In circuit-switched networks, the QoS is formed by several technical metrics: reliability,
transmission quality, stability of service quality etc. In particular, it is possible to provide the
level of availability called 99.999 percent, and to offer a lower value of availability for
customer that do not need the top QoS, saving money for network services.


- Availability
To achieve the highest levels of network availability, networks have to implement network
recovery techniques at both the data and the transport layers. These restoration schemes
are designed to protect the network from all types of failures including cable cuts, router
card failures, and power outages.
Availability of connections is defined per SLA by means of different resilience mechanisms,
with short recovery times.


       7.3.57.3.4 Traffic Management Requirements

- Network architecture scalable and flexible in order to deal with a traffic volume
increasing and traffic variations
The interactions between the IP-layer functionality of packet networks and that of the
optical transport layer, requires building in end-to-end efficiencies, provisioning services
quickly, and, in particular, providing services based on real-time traffic patterns.


- Bandwidth adjustment capability
The service provider can also set-up the service where the network dynamically and
automatically increases/decreases bandwidth as traffic volumes/patterns change. If the
demand for bandwidth increases unexpectedly, additional bandwidth can be dynamically
provisioned for that connection. This includes overflow bandwidth or bandwidth over the
stated contract amount. The triggering parameters may be utilization thresholds, time-of-
day, day-of-month, per-application volumes, etc.


- Traffic Engineering Provisioning
To achieve ever-greater efficiencies, optical service providers must streamline their
operations by reducing the number of people required to deliver these services, and



                                        Page 114 of 168
                             IST IP NOBEL "Next generation
                                                                                Title of the document
                             Optical network for Broadband
                                  European Leadership"                    e1686d81-2994-497c-93f6-
                                                                   b2e913b07f59.docD06_DRAFT_V1
                                                                               6JW2 Andreas zweiter
                                                                        Teil.docD06_DRAFT_V15ejw




reducing the time required to activate and troubleshoot network problems. To accomplish
these objectives, they are focusing on automated provisioning.
GMPLS is designed to enable multi-vendor and multi-layer provisioning. Therefore,
requests for services in the data network that may require connectivity or reconfigurations
at the optical layer can happen in a more automated fashion. In addition, instead of
provisioning on a site-by-site basis, GMPLS creates a homogenous network where
provisioning is performed network-wide.


- General Traffic Engineering Requirements
Traffic engineering is defined as that aspect of network engineering dealing with the issue
of performance evaluation and performance optimization of operational networks. Traffic
Engineering encompasses the application of technology and scientific principles to the
measurement, characterization, modelingmodelling, and control of Internet traffic (RFC
2702). An important objective of traffic engineering is to facilitate reliable network
operations. These operations can be facilitated by providing mechanisms that enhance
network integrity and by embracing policies emphasizing network survivability. This results
in a minimization of the vulnerability of the network to service outages arising from errors,
faults, and failures occurring within the network.
At first this section discusses general traffic management requirements and instruments,
which are independent from the network layer (Ll1-Ll3). The following list of instruments
may be applicable to the solution context of traffic engineering:
(1) A set of policies, objectives, and requirements (which may be context dependent) for
network performance evaluation and performance optimizationoptimisation.
(2) A collection of online and possibly offline tools and mechanisms for measurement,
characterization, modelingmodelling, and control of Internet traffic and control over the
placement and allocation of network resources, as well as control over the mapping or
distribution of traffic onto the infrastructure.
(3) A set of constraints on the operating environment, the network protocols, and the traffic
engineering system itself.
(4) A set of quantitative and qualitative techniques and methodologies for abstracting,
formulating, and solving traffic engineeringtraffic-engineering problems.
(5) A set of administrative control parameters, which may be manipulated through a
Configuration Management (CM) system. The CM system itself may include a configuration
control subsystem, a configuration repository, a configuration accounting subsystem, and a
configuration auditing subsystem.
(6) A set of guidelines for network performance                   evaluation,   performance
optimizationoptimisation, and performance improvement.
Furthermore the four phases of the TE process model described below should be
considered in core as well as in metropolitan networks [RFC 2702]:
   -   Definition of a relevant control policy that governgoverns the operation of the
       network (depends on many factors (business model, network cost structure,
       operating constraints, etc.)




                                      Page 115 of 168
                             IST IP NOBEL "Next generation
                                                                               Title of the document
                             Optical network for Broadband
                                  European Leadership"                    e1686d81-2994-497c-93f6-
                                                                   b2e913b07f59.docD06_DRAFT_V1
                                                                               6JW2 Andreas zweiter
                                                                        Teil.docD06_DRAFT_V15ejw




   -   Feedback mechanism involving the acquisition of measurement data from the
       operational network
   -   Analyzing the network state and to characterize traffic workload. Performance
       analysis may be proactive and/or reactive
   -   Performance optimization of the network. The performance optimization phase
       involves a decision process, which selects and implements a set of actions from a
       set of alternatives.


- Multilayer traffic engineering support
Traffic Engineering (TE) is concerned with performance optimisation of operational
networks. In general, it encompasses the application of technology and scientific principles
to the measurement, modelling, characterization, and control of Internet traffic, and the
application of such knowledge and techniques to achieve specific performance objectives.
One of the major goals of TE is to facilitate efficient and reliable network operations while
simultaneously optimising network resource utilization and traffic performance. TE has
become an indispensable function in many large Autonomous Systems because of the
high cost of network assets and the commercial and competitive nature of the Internet.
These factors emphasize the need for maximal operational efficiency. TE is defined as that
aspect of network engineering dealing with the issue of performance evaluation and
performance optimisation of operational networks.
Using the word "multilayer" means here that we are considering a network where two (or
more) levels inter-works. In particular, the network layers may be IP and optical transport
adopting ASON/GMPLS technologies.


- Multiregion TE support
Multi-region TE support is a furthermore requirement for optical core and metro networks.
Multi-region traffic engineering in GMPLS networks increases network resource efficiency,
because all the network resources are taken into account at the same time. However, in
GMPLS multi-region network environments, traffic engineering becomes more complicated,
compared with that in single-region network environments. A solution, which meets this
requirement, is described in [draft-oki-ccamp-gtep-00.txt]
This proposal describes a generalized traffic engineering protocol (GTEP). GTEP is a
protocol that communicates between a Constrained Shortest Path First (CSPF) engine and
a GMPLS controller (GMPLS CNTL). A GMPLS CNTL is a controller that handles GMPLS
and MPLS protocols such as routing and signalingsignalling protocols and controls a
GMPLS node. The CSPF engine provides a function of traffic engineering, which calculates
LSP routes and judges whether a new lower-layer LSP should be established.
Basically multi-region TE must support the following functions:
      Inter-AS Bandwidth Guarantees
      Inter-AS Resource Optimization
      Fast Recovery across ASes




                                      Page 116 of 168
                               IST IP NOBEL "Next generation
                                                                                    Title of the document
                               Optical network for Broadband
                                    European Leadership"                      e1686d81-2994-497c-93f6-
                                                                       b2e913b07f59.docD06_DRAFT_V1
                                                                                   6JW2 Andreas zweiter
                                                                            Teil.docD06_DRAFT_V15ejw




       7.3.67.3.5 Security and Authentication Requirements

- Provide security from network operator and user perspective
Since optical connections carry high volumes of data, involve multiple stakeholders, and
consume significant network resources, security mechanisms are required to safeguard a
transport network against attacks on the control plane, on the management plane, and
unauthorized use of network resources in the data plane.


- Allow for operators to limit the visibility of the network topology to authorized
entities
Operator may limit the visibility of their network information in the form, for instance, of
some routing messages, which are exchanged between different network domains. While
reachability information is distributed generally to neighbouring networks, node and link
topology information may be kept invisible and is not advertised to other adjacent networks.


- Be able to verify the identity of users and terminals. Additionally it shall be able to
check the authorization of the user to use resources of the NGN and to access
services offered by an NGN.
The network should provide mechanisms to authenticate entities exchanging information
across an interface, to guarantee the integrity of the information (data integrity) exchanged
across an interface (neighbour discovery, service discovery, topology and resource status,
signalling, etc.) and to protect the confidentiality of certain types of information. These
mechanisms should protect against both malicious attacks against the network as well as
unintentionally malfunctioning control entities.


- Identities of NGN users, used e.g. for authentication, authorization, and routing,
shall be administered by the network operator and shall not be changeable by the
user.
Since the relationship between the network operator that controls the optical layer and the
one providing client connectivity is often a billable one, UNI is very concerned with issues
like client identification and authentication, discovery of service availability within the optical
layer, and how to signal various service parameters such as bandwidth, class of service for
restoration, and characteristics of the data such as SONET/SDH port or concatenation
information. Authentication, authorization, and routing are typical functionality, for which the
network operator is responsible in the network and may not be impacted from client
networks without express permission.
For instance, because the topology of the optical cloud may remain hidden to clients, there
can be no explicit route set up; instead UNI identifies the source and destination nodes
using an address that is assigned to each called the Transport Network Assigned (TNA)
address.




                                        Page 117 of 168
                             IST IP NOBEL "Next generation
                                                                                 Title of the document
                             Optical network for Broadband
                                  European Leadership"                     e1686d81-2994-497c-93f6-
                                                                    b2e913b07f59.docD06_DRAFT_V1
                                                                                6JW2 Andreas zweiter
                                                                         Teil.docD06_DRAFT_V15ejw




          7.3.77.3.6 Addressing requirements

- Both telecom and Internet numbering and addressing schemes shall be supported
as user identities.
An addressing scheme is required at the optical network layer. As optical transport
networks become wide spread, it is foreseen that an addressing hierarchy may be
required. Optical layer addressing also requires address resolution protocols for higher
levels to communicate across optical domains.


          7.3.87.3.7 Billing and charging Requirements

- Differences in terminology between “Telecommunications World” and “IP-World”
At first this section starts with clarification the terminology between “Telecommunications
World” and “IP-World”. It is, however, easy to note that the different working groups do not
always work closely together, or that they do not share the same business model as the
basis of their work. As a result, several mismatches have been identified in the specified
functionality and even in the adopted terminology.


Terms        Definition in the Telecommunications          Definition in the IP-World
             World
Charging     The functions whereby information related     A function that derives non-
             to a chargeable event is formatted and        monetary cost for accounting data
             transferred in order to make it possible to   sets based on service and
             determine usage for which the charged         customer-specific tariftariff
             party may be billed.                          parameters.
Billing      The functions whereby charging data are       A function that translates costs
             transformed into bills requiring payment.     calculated by the charging into
                                                           monetary units and generates a
                                                           final bill for the customer.
Accounting The process of apportioning charges             A function that describes the col-
           between the home environment, serving           lection of data about resource
           network, and user.                              consumption. This includes the
                                                           control of data gathering, trans-
                                                           port, & storage of accounting data.

Table 323232: Terminology.
For example, Table 32:Table 32:Table 32: presents the differences in the terminology used
by the two worlds in relation to charging, billing, and accounting. For instance, the IETF’s
“Accounting” functionality can be mapped to the telecommunications industry’s “Charging”
functionality. Moreover, the telecommunications industry’s “Accounting” term is not used at
all by the IETF since it is out of the scope of their work.




                                      Page 118 of 168
                              IST IP NOBEL "Next generation
                                                                                 Title of the document
                              Optical network for Broadband
                                   European Leadership"                    e1686d81-2994-497c-93f6-
                                                                    b2e913b07f59.docD06_DRAFT_V1
                                                                                6JW2 Andreas zweiter
                                                                         Teil.docD06_DRAFT_V15ejw




- Charging Provisioning
During the past several yearsyears’ application/service providers have exploited the
explosive growth of the Internet to offer their services to end-users. The flat-rate model that
has been adopted to charge people for accessing the network was a simple one and did
not require complex systems for monitoring and billing purposes. Application/service
providers’ revenue, were based mainly on advertisements, since their services and content
were usually offered to users free of charge.
However, the development and introduction of advanced services led to new requirements
on the part of users, ISPs, application/service providers and network providers. The
following points describing requirements from a different perspective.
        Service Provider´s should account or charge based on the content and service
         usage.
        Customer’s want shift from best effort to QoS model.
        ISP´s need new accounting models that take into account the utilization and sharing
         of network resources in order to apply efficient network management schemes
         (waste of bandwidth).
        Network providers should be able to meter network traffic and consumption of
         network resources (including obviously Layer 1 resources).




       Figure 373838: Functional model for resource usage and billing processing.
Figure 37Figure 38Figure 38 presents the main functional model for collecting and
processing information related to resource usage and billing of users [RFC 2975].
Network and service providers must support the following functions of collecting and
processing information:
        Metering function
In this model, metering is the function of capturing all data related to network resources’
consumption (e.g., volume of exchanged data) and is performed by network devices. To
achieve this functionality there is a need for the network devices to export flow information
in a standardized way. An IP flow information export system includes a data model, which
represents the flow information, and a transport protocol.




                                      Page 119 of 168
                                     IST IP NOBEL "Next generation
                                                                                               Title of the document
                                     Optical network for Broadband
                                          European Leadership"                            e1686d81-2994-497c-93f6-
                                                                                   b2e913b07f59.docD06_DRAFT_V1
                                                                                               6JW2 Andreas zweiter
                                                                                        Teil.docD06_DRAFT_V15ejw




This metering function must be supported by all network elements including Layer 1-Layer3
interfaces. The collection of metering data can be initiated either by the network device
itself (push model), or by the accounting server, which plays the role of the collector entity
(pull model).
     Acconting function
The accounting function, which is performed by the accounting server, is responsible for
the collection and storage of the accounting data. Accounting may also include
summarization of interim information, elimination of duplicated data, and generation and
processing of session records. Moreover, session records and their related IDs are also
produced and handled. Accounting is also responsible for forwarding these records to other
peer entities in the case of roaming terminals. The accounting attributes depend on the
applied charging model (e.g., flat-rate, session-oriented, time-based, volume-based, etc.).
        Billing function
The billing function deals with bill preparation and presentation to the party that is
responsible for payment. This function receives session records or processed accounting
data from the accounting function via a transfer protocol such as SMTP, FTP, or HTTP.
Finally, it prepares an invoice according to the appropriate billing policy (e.g., computing a
special discount for a user, addition of a monthly fee, etc.).
More detailed requirements depending on the business model of the network or service
provider (tariff model). The following two examples describe the different accounting
requirements.


- Accounting requirements (Settlements between NPs)
In the tariff model between NPs the ingress NP pays for the traffic sent to the egress NP.


                            NPingress                              NPTGN                        NPegress




                    LER            ASBRingress            ASBR1iGN     ASBR2iGN              ASBRegress

 CPESP
                                                              transit traffic
                               Cingress            Y1              CTGN               Z1
                                                          in affi
                                                            te c
                                                             tr

                                                              r-
                                                                 do
                                                                   m
                                                                     ai
                                                                       n




                                                                   LER                            LER
                             LER




                                   CPESP                                   CPESP                        CPESP


                            Figure 383939: tariff model between NPs.




                                                 Page 120 of 168
                                                   IST IP NOBEL "Next generation
                                                                                                             Title of the document
                                                   Optical network for Broadband
                                                        European Leadership"                            e1686d81-2994-497c-93f6-
                                                                                                 b2e913b07f59.docD06_DRAFT_V1
                                                                                                             6JW2 Andreas zweiter
                                                                                                      Teil.docD06_DRAFT_V15ejw




The tariff model for NPTGN looks like for each CoS as follows

-     NPTGN pays Z1 (e.g. Z1 €/byte) for transit traffic sent to NPegress and

-     NPTGN gets Y1 for transit traffic received from NPingress.
It is the goal of NPTGN to ensure that Y1 = Z1 + CTGN with CTGN representing his transit costs
(e.g. the costs for the network infrastructure). Thus, NPTGN must exactly know the amount
of transit traffic per CoS caused by NPingres and transmitted to NPegress in order to calculate
the tariff for NPingress. NPTGN must be able to distinguish between transit traffic and inter-
domain traffic terminating in its own network. Various tariffs might be raised by the different
egress NPs. This requires CoS and destination based accounting mechanisms on
ASBR1TGN for the received traffic from NPingress. It is not intended to introduce a distance
based tariff model.
Accounting Requirements (Settlements between SPs and NPs)
The goal of the settlement between SPs and NPs is the assignment of the costs between
the NPs to each SP. The SP’s tariff model for the transit traffic looks like for each CoS as
follows: X1 = Y1 + Cingress with Cingress representing the internal costs of NPingress. This
requires CoS and destination based accounting mechanisms on the ingress LERs for the
sent traffic in order to be able to distinguish between intra-domain and transit/inter-domain
traffic. It is not intended to introduce a distance based tariff model.
Note, intra-LER traffic must also be accounted. In order to avoid double billing the SP has
either to pay for the sent or for the received traffic.


                                          NPingress                              NPTGN                          NPegress


                 intra-LER traffic
    CPESP
                          LER                    ASBRingress            ASBR1iGN     ASBR2iGN              ASBRegress



    CPESP
                                                                            transit traffic
            X1                               Cingress            Y1              CTGN               Z1
                                                                        in affi
                                                                          te c




                       intr
                                                                           tr

                                                                            r-d




                            a-d
                                                                               om




            X2                 om
                                 ain
                                                                                   ai




                         C
                                                                                     n




                          ing        t       raf
                                res              fic
                                      s                                          LER                            LER
                                           LER




                                                 CPESP                                   CPESP                        CPESP


                                          Figure 394040: Tariff model between NPs.




                                                               Page 121 of 168
                             IST IP NOBEL "Next generation
                                                                               Title of the document
                             Optical network for Broadband
                                  European Leadership"                    e1686d81-2994-497c-93f6-
                                                                   b2e913b07f59.docD06_DRAFT_V1
                                                                               6JW2 Andreas zweiter
                                                                        Teil.docD06_DRAFT_V15ejw




       7.3.97.3.8 Criteria for network evolution requirements

- Multi-layer integrated data, control, and management plane
In the last decade Network Providers largely deployed Sonet/SDH technology in
core/metro transport networks. Currently network requirements for the evolution of
transport network are driven by a few elementary needs, such as: new network solutions
have to optimise the use of resources (reducing CAPEX); have to reduce the operating
costs (reducing OPEX); have to improve quality, efficiency in providing current and new
services (increasing and generating new revenues).
Furthermore, given the investments made in the last few years, traffic demand is growing
today less-than-expected per year (gated more by the rate of adoption of new
applications/services): as such another major problem for the evolution of transport
network seems to be that there is not enough demand to deploy more capacity at the
current costs (a chicken-egg problem).
To achieve the above mentionedabove-mentioned goals (reducing CAPEX, OPEX,
generating new revenues), ASON/GMPLS solutions have the flexibility to allow different
levels of integration of the network functionality (e.g. L3 routing, and L1-L2 switching) and
different levels control/management integration within the network elements. As such two
related basic questions to be considered defining network evolution requirements are:
1. What are network positioning (what PoPs metro/core) and integration level of L3-L2-L1
   functionality as a function of time in the network evolution?
2. What's the integration level of the network/node control as a function of time in the
   network evolution?
Even if these questions, undoubtedly, depends upon Providers' individual strategic
decisions, general techno-economic criteria can be defined. The questions are very much
related but former seems to impact more on CAPEX (as more related to network design,
planning, and dimensioning), the latter is likely to impact more on OPEX (as more related
to automatic provisioning, discovery, inventory).
The emerging of the ASON/GMPLS technology and its potential introduction in the
Transport Network (metro/core) creates the problem of how to evolve from the current
legacy network to the new one.
Unless the Network Provider is building a completely new network, this problem depends
on the strategic decisions about the type of network equipments and solutions are to be
introduced and in what evolutionary paths thus satisfying the network evolution
requirements.
The above decisions, such as the time order in which different types of equipment are
introduced and the type of interworking used, will affect subsequent stages of evolution and
may pose networking or interworking constraints.
Appendix III of ITU-T Recommendation G.872 provides information on how a transport
network could evolve to one based on the Optical Transport Network. The same basic
approach can be adopted in defining how a transport network could evolve to one based
on the NobelNOBEL investigations.




                                      Page 122 of 168
                              IST IP NOBEL "Next generation
                                                                                   Title of the document
                              Optical network for Broadband
                                   European Leadership"                     e1686d81-2994-497c-93f6-
                                                                     b2e913b07f59.docD06_DRAFT_V1
                                                                                 6JW2 Andreas zweiter
                                                                          Teil.docD06_DRAFT_V15ejw




For example, the two most important questions are identifying the types of client layer
networks to be transported by the evolving transport network and the type of interworking.
The definition of the network evolution requirements is a very complex problem whose
basic motivations are reducing CAPEX, OPEX and increasing, generating new revenues.
Once some basic scenarios are identified, the types of client layer networks to be
transported and the type of interworking (old versus new core/metro) decided will dictate
the constraints for the technology introduction and the network evolutionary paths.


       7.3.107.3.9 Control plane requirements

- Unified control plane
As Telecommunications carriers look to evolve their existing networks or to build a new
integrated network infrastructure, a fundamental paradigm shift is emerging in the way
networks are designed. From a layered network model involving the management of
network elements individually at each layer, the philosophy is shifting to one of an
integrated infrastructure, where one can seamlessly manage packets, circuits, and light
paths. The reasons for this industry trend towards a unified set of mechanisms (the unified
control plane) that will enable service providers to manage separate network elements in a
uniform way can be traced to the historical evolution of transport and packet networks. The
Internet Protocol (IP) and its associated protocols matured to become the ubiquitous
routing and signalling mechanisms for transporting all kind of data. The IP-based GMPLS
provides a single, unified control plane for multiple switching layers.

- Backward compatibility in order to facilitate network migration
Backward compatibility with non-proprietary control planes is illustrated by the following
example. Backward compatibility allows for a migration or coexistence with GMPLS RSVP-
TE use. ASON requires that for any new and existing GMPLS features, transit nodes do                   Comment [POA5]: Seems to be a very
not need to be updated and do not need to modify their behaviour to support the end-to-                restricted set of backward compatibility – only
                                                                                                       with older types of control planes? What about
end features of ASON.                                                                                  non-control-plane networks? This is after all the
                                                                                                       second paragraph from top on “network
                                                                                                       evolution” (6.3.8) – should it not start with
                                                                                                       something wider, like SDH… and why is this not
- Vertical integration                                                                                 under “control plane requirements…”

The integrated model encompassing several data planes, a vertically integrated control
plane and the corresponding management plane. Furthermore, in multi-switching
environment, vertical integration for the control plane refers to the definition of collaborative
mechanisms within a single control plane instance driving multiple (but at least two) data
planes (also referred in the scope of GMPLS as switching layers).
In addition, from a layered network model involving the management of network elements
individually at each layer, the philosophy is shifting to one of an integrated infrastructure,
where one can seamlessly manage packets, circuits, and lightpaths.


- Provisioning of end-to-end connections over entire network. Requirements for
  inter-domain routing, inter-domain interface definition, GMPLS for all domains




                                       Page 123 of 168
                              IST IP NOBEL "Next generation
                                                                                  Title of the document
                              Optical network for Broadband
                                   European Leadership"                     e1686d81-2994-497c-93f6-
                                                                     b2e913b07f59.docD06_DRAFT_V1
                                                                                 6JW2 Andreas zweiter
                                                                          Teil.docD06_DRAFT_V15ejw




The control plane must have the capability to establish, teardown, and maintain the end-to-
end connection, and the hop-by-hop connection segments between any two end-points.
With reference to the transport layers, higher layers generate connection requests with
specific attributes like bandwidth, quality, policy/priority, and survivability. Such attributes
specify the optical Quality of Service requirements like path bandwidth, delay, and BER.


- Unified inter-domain control (horizontal integration)
The domain interconnection or boundary problem at control plane level may be expressed
as the horizontal integration issue. Horizontal integration is defined when each entity
constituting the network environment includes at least one common (data plane) switching
capability and the control plane topology extends over several partitions, being areas. In
the latter case, the integration is thus defined between nodes hosting the same switching
capability. For instance, the control plane interconnection between lambda switching
capable areas defines a horizontal integration.
In this context, the networks - including the control plane - offer the necessary inter-domain
interfaces as required between users/provider and provider/provider. Ever since the kind of
networks we are studying consists of three main levels (management, control, and data
plane), it is very important that the inter-domain interoperability must be considered over all
these three points of view.


- Control Plane and NMS robust against signalling and management network failure
  and able to preserve from traffic disruption
The control plane is considered a managed entity within a network. Therefore, it is subject
to management requirements just as other managed entities in the network are subject to
such requirements. Today’s optical network architectures lack the proper control
mechanisms that would interact with the management layer to provide fast network
reconfiguration. However, both the control and management plane should be robust
enough against failures, and assuring the preservation of traffic continuity.
- Network auto discovery & Control Plane Resilience
With automated network discovery, the network autonomously discovers new equipment or
changes to existing equipment. Additionally, the control plane automates the capacity
assessment and path computation process. These capabilities substantially reduce the
amount of time it takes service providers to provision services and make network changes.
Furthermore, as capacity and/or equipment become available and as customers cancel,
disconnect, or change orders, resources can be readily made available to other customers.
Without optical signalling, discovering available resources and "actually" making them
available is a painful process for most service providers.
With control plane resilience, the network element can discover the existing cross-connects
after recovering from the control plane failure. For example, when only control plane failure
occurs within one network element, the optical cross-connects will still be in place carrying
data traffic. After recovery of the control plane, the network element should automatically
assess the data plane (i.e. optical cross-connects), and reconfigure its control plane so that
it can synchronizesynchronise with other control plane entities.




                                       Page 124 of 168
                             IST IP NOBEL "Next generation
                                                                               Title of the document
                             Optical network for Broadband
                                  European Leadership"                    e1686d81-2994-497c-93f6-
                                                                   b2e913b07f59.docD06_DRAFT_V1
                                                                               6JW2 Andreas zweiter
                                                                        Teil.docD06_DRAFT_V15ejw




- Appropriate network visibility among different administrative domains belonging to
  different operators
Administrative domains may have multiple points of inter-connections. All relevant interface
functions, such as routing, reachability information exchanges, and inter-connection
topology discovery must be recognizedrecognised at the interfaces between those
domains. The control plane should provide the mechanisms to form the appropriate
visibility among different administrative domains.


- Fast Provisioning
As part of the reliable optical network design, fast provisioning of optical network
connections contributes to the efficient service delivery, and OPEX reduction, and helps
reaching new customers with broadband services.


- Automatic provisioning
To achieve ever-greater efficiencies, optical service providers must streamline their
operations by reducing the number of people required to deliver these services, and
reducing the time required to activate and to troubleshoot network problems. To
accomplish these objectives, they are focusing on automated provisioning through a
distributed control plane, which is designed to enable multi-vendor and multi-layer
provisioning in an automated way. Therefore, requests for services in the data network that
may require connectivity or reconfigurations at the optical layer can happen in a more
automated fashion. In addition, instead of provisioning on a site-by-site basis, the control
plane creates a homogenous network where provisioning is performed network-wide.


- Towards bandwidth on demand services
The service provider can also set-up the service where the network dynamically and
automatically increases/decreases bandwidth as traffic volumes/patterns change. If the
demand for bandwidth increases unexpectedly, additional bandwidth can be dynamically
provisioned for that connection. This includes overflow bandwidth or bandwidth over the
stated contract amount. The triggering parameters may be utilizationutilisation thresholds,
time-of-day, day-of-month, per-application volumes, etc.
Bandwidth on demand provides connectivity between two access points in a non-pre-
planned, fast, and automatic way using signalling. This means also dynamic reconfiguring
of the data carrying capacity within the network and that restoration is also considered here
to be a bandwidth on demand service.
In this environment, the service provider can set-up a service where the network
dynamically and automatically increases/decreases bandwidth as traffic volumes/patterns
change. If the demand for bandwidth increases unexpectedly, additional bandwidth can be
dynamically provisioned for that connection. This includes overflow bandwidth or bandwidth
over the stated contract amount. The triggering parameters may be utilizationutilisation
thresholds, time-of-day, day-of-month, per-application volumes, etc.




                                      Page 125 of 168
                              IST IP NOBEL "Next generation
                                                                                 Title of the document
                              Optical network for Broadband
                                   European Leadership"                    e1686d81-2994-497c-93f6-
                                                                    b2e913b07f59.docD06_DRAFT_V1
                                                                                6JW2 Andreas zweiter
                                                                         Teil.docD06_DRAFT_V15ejw




- Flexibility – re-configurable transport/optical layer
A network operator may have many reasons for wanting to reconfigure the network,
primarily motivated by who is paying for what. Flexibility of the transport layers means a fair
allocation of bandwidth between competing routes dealing with bursts of activity over many
timescales. Re-configurability increases network flexibility and responsiveness to dynamic
traffic demands/changes.


       7.3.117.3.10 Interoperability and Interworking Requirements

- Control plane interworking (UNI/NNI)
The ASON model distinguishes reference points (representing points of information
exchange) defined (1) between a user (service requester) and a service provider control
domain a.k.a. user-network interface (UNI), (2) between control domains a.k.a. external
network-network interface (E-NNI) and, (3) within a control domain a.k.a. internal network-
network interface (I-NNI). The I-NNI and E-NNI interfaces are between protocol controllers,
and may or may not use transport plane (physical) links. It must not be assumed that there
is a one-to-one relationship of control plane interfaces and transport plane (physical) links,
or that there is a one-to-one relationship of control plane entities and transport plane
entities, or that there is a one-to-one relationship of control plane identifiers for transport
plane resources.


- Multi-domain Interoperability
In many of today's complex networks, it is impossible to engineer end-to-end efficiencies in
multi-domain environment, provision services quickly, or provide services based on real-
time traffic patterns without the ability to manage the interactions between the IP-layer
functionality of packet networks and that of the optical layer. According to proponents of
ASON/GMPLS, an optical control plane is the most advanced and far-reaching means of
control these interactions.
Other issue arises is that of translating resilience classes from one domain to another.
I-NNI /E-NNI can resolve that issue.


- Multi-vendor interoperability
The multi-vendor interoperability of metro and core solutions maximizesmaximises, in
particular, carrier performance, and ensures the interoperability with legacy and emerging
network architectures. One of the most important objectives of the development of a
standardizedstandardised ASON/GMPLS control plane is to contribute to interoperability,
which validate the speed and ease of provisioning enabled by ASON/GMPLS in a live,
multi-vendor network.


- Seamless Boundary in between Networks




                                      Page 126 of 168
                              IST IP NOBEL "Next generation
                                                                                   Title of the document
                              Optical network for Broadband
                                   European Leadership"                     e1686d81-2994-497c-93f6-
                                                                     b2e913b07f59.docD06_DRAFT_V1
                                                                                 6JW2 Andreas zweiter
                                                                          Teil.docD06_DRAFT_V15ejw




Given the vast amount of legacy SONET/SDH equipment, there is a clear need for an
efficient inter-working between traditional circuit oriented networks and IP networks based
on the packet switching paradigm. For example, efficient control plane interworking
between the IP/MPLS and SONET/SDH GMPLS layers is indispensable and requires the
specification of their coordination.
Control plane interconnection models with different degrees of interaction can be defined,
with the overlay model and the peer model as the extremes and the augmented model as
intermediate. Seamless boundary between networks is related to the signalingsignalling
and routing capabilities embedded into UNI, E-NNI, and I-NNI.



       7.3.127.3.11 Management plane requirements

- Easy to use Network
Emerging standards and technologies for optical networks allow for a significantly
simplified architecture, easy and quick to provision services, more effective management,
better interoperability and integration, and overall lower cost. In addition, it will be possible
to provision services on these future networks such that global applications will be able to
be much more location independent.


- Transparent for applications - hide network technology to users
There are multiple separate service, technology and technical considerations for networks
depending on location, at the metro edge, metro core, aggregation points, long haul, and
ultra long haul. Next generation optical networking has a potential to significantly reduce or
eliminate all of these barriers and especially with regard to application and end users. To
some degree, one of the key goals in this development is to create network services with a
high degree of transparency, that is, allow network technical elements to become invisible
while providing precise levels of required resources to applications and services.


- Monitoring of end-to-end QoS & QoR
The requirement of integrated monitoring of the (optical) performance of connections, QoS,
and fault management speed system installation and wavelength turn-up and simplifies
ongoing maintenance. Furthermore, the management plane should be able to monitor end-
to-end Quality of Resilience (QoR). That means, the end-to-end type of transport plane
resilience parameters (such as recovery time, etc) be monitored and adhered according to
the SLAs.


- Connectivity and network performance supervision
As networks run faster and become more complex, infrastructure, links, and devices must
operate to precise levels in a tighter performance. As a result, a huge number of network




                                       Page 127 of 168
                              IST IP NOBEL "Next generation
                                                                                Title of the document
                              Optical network for Broadband
                                   European Leadership"                   e1686d81-2994-497c-93f6-
                                                                   b2e913b07f59.docD06_DRAFT_V1
                                                                               6JW2 Andreas zweiter
                                                                        Teil.docD06_DRAFT_V15ejw




problems stem from simple wiring and connection problems. Connectivity and performance
supervision is at the heart of the efficient network management.


- Network Monitoring
A monitoring system is dedicated to the supervision of the physical and optical layers of a
network. Optical-layer monitoring should provide valuable, accurate information about the
deterioration or drift of slow- and small-variation signals, helping to detect problems before
they affect the quality of service (QoS). It helps maintain the system from a lower layer’s
perspective.


- Policy-based management (network and local-basis)
Today’s optical network architectures lack the proper control mechanisms that would
interact with the management layer to provide fast reconfiguration. The problem of
accurate intra-domain provisioning in an automated manner allows satisfying the contracts
with customers while optimising the use of the network resources. It is required that the
policy-based management system dynamically guides the behaviour of such an automated
provisioning through the control plane in order to be able to meet the high-level business
objectives. Therefore, the emerging policy-based management paradigm is the adequate
means to achieve this requirement.


- End-to-end traffic management (CAC, bandwidth management, Policing)
Traffic management features are designed to minimisze congestion while maximiszing the
efficiency of traffic. Applications have precise service requirements on throughput,
maximum delay, variance of delays, loss probability etc. The network has to guarantee the
required Quality of Service. For instance, the primary function of the Connection Admission
Control (CAC) is to accept a new connection request only if it’s stated QoS can be
maintained without influencing the QoS of the already accepted connections. Traffic
management features are key elements in efficient networking.


- Multi-vendor interoperability
In the near future, network element management interfaces and OSS interfaces will be pre-
integrated by control plane vendors. Indeed, independent control planes increase the
performance of network elements and OSS, and reduce the carriers’ reliance on any single
network element or OSS application. This eliminates the task of integrating new network
elements into myriad OSS applications.


 7.4     General

       7.4.1 Open Networks
The network should be available for all service providers, allowing service/application
providers to offer services to the individual end-users. . Each user should have the



                                      Page 128 of 168
                              IST IP NOBEL "Next generation
                                                                              Title of the document
                              Optical network for Broadband
                                   European Leadership"                  e1686d81-2994-497c-93f6-
                                                                  b2e913b07f59.docD06_DRAFT_V1
                                                                              6JW2 Andreas zweiter
                                                                       Teil.docD06_DRAFT_V15ejw




opportunity to selectively choose between different services and service providers. This
requirement together with bottleneck., These two requirements – neutrality and compteting
comptetioncompetition among service providers - lead to the concept of an open network.


- Open networking
An open interface is, in this sense, an interface where business can be made effective. To
get a reliable business operation with low OPEXpex, that attracts many customers, the
interface must be:
      Specifiable
      Verifiable
      Reliable
Otherwise, the business will not be effective.


- Competition between application providers


       7.4.2 Price reduction on transport

- Cheap advanced optical transmission
The basic optical network layer that delivers its transport capacity at price-levels of 10-5
Euro/bit must still have an availability of 99.999% - standard telecom quality. Individual
services running on top of this network may have varying performance.
Feasibility of a network depends mainly on its cost effectiveness, which can be achieved by
proper control and management of the resources.




                                      Page 129 of 168
                              IST IP NOBEL "Next generation
                                                                               Title of the document
                              Optical network for Broadband
                                   European Leadership"                   e1686d81-2994-497c-93f6-
                                                                   b2e913b07f59.docD06_DRAFT_V1
                                                                               6JW2 Andreas zweiter
                                                                        Teil.docD06_DRAFT_V15ejw




 7.5     Service Requirements




       Figure 404141: Various service categories and distinct groups of players.


       7.5.1 General
Apply the dynamic configuration to the transport layer(s).


- Provide at least following service capabilities

The network should provide at least the following service types:
    Authorization & Authentication
       Offline & online charging
       Location
       Policy control
       Session handling
       Provision of the bearer(s)
       Message based information exchange
Whilst ever users have fears about security issues, they will be reluctant to use certain
services. Intranet security is a fundamental part of any enterprise network design. While
every enterprise network has valuable information to protect, authorizedauthorised users
still need access. For many years, network managers deployed closed data centres or
separate networks to provide Intranet security, but widely used mission-critical applications
and network consolidation call for more security.




                                      Page 130 of 168
                             IST IP NOBEL "Next generation
                                                                               Title of the document
                             Optical network for Broadband
                                  European Leadership"                    e1686d81-2994-497c-93f6-
                                                                   b2e913b07f59.docD06_DRAFT_V1
                                                                               6JW2 Andreas zweiter
                                                                        Teil.docD06_DRAFT_V15ejw




- Support services like
The network should support the following services:
      voice
      multimedia (audio & video, incl. streaming)
      Real Time Cinema services
      Service – capability to support emerging multiparty communication such as
       conferencing messaging services gaming
      Internet access
      Bandwidth on demand


- Local Network Mirrors
Local network mirrors can be viewed as the server mobility approach, where the server is
changing in time with reference, for instance, to cost constraints requested by the
application or user. In this context, server means any kind of resources located outside of
the network. In particular, attractive multimedia services are concerned with local network
mirrors.


- Bandwidth for Sale
The optical bandwidth cost is continuously decreasing in such a way that it could substitute
today for all the electronic processing. The cost evolution of bandwidth indicates that local
and interexchange carrier cost could dramatically fall. Carriers, resellers, and service
providers are counting on escalating demand in bandwidth, especially, from large
organizationsorganisations and academia, which can trade for big commitments in
bandwidth usage.


- Support peer-to-peer and client-server communication types
Peer-to-peer is a communications model in which each party has the same capabilities and
either party can initiate a communication session. Other models with which it might be
contrasted include the client/server model. In some cases, peer-to-peer communications is
implemented by giving each communication node both server and client capabilities. In
recent usage, peer-to-peer has come to describe applications in which users can use the
Internet to exchange files with each other directly or through a mediating server. The
client/server model requires a server to store data that is retrieved by the clients.


- An architectural framework shall be provided by the NGN that enables maximum
flexibility in the end user devices and network servers incl. application servers.




                                      Page 131 of 168
                              IST IP NOBEL "Next generation
                                                                                Title of the document
                              Optical network for Broadband
                                   European Leadership"                   e1686d81-2994-497c-93f6-
                                                                   b2e913b07f59.docD06_DRAFT_V1
                                                                               6JW2 Andreas zweiter
                                                                        Teil.docD06_DRAFT_V15ejw




This framework shall enable an operator to efficiently deploy IP multimedia applications
without having to wait for these applications or additional enabling technology to be
standardizedstandardised.


- Provisioning of end-to-end transparent transport services
High speeds for performing certain functions, high availability of the system to perform
these functions, and accuracy of the information transferred make part of users’ QoS
requirements. These translate into network performance values, for instance, for
availability, bandwidth/throughput, end-to-end delay, end-to-end jitter, and bit/packet loss.
The requirements are end-to-end transparent requirements in the sense that the users are
concerned with the behaviour of the system at the points where they interact with the
system, not with what happens inside.


       7.5.2 QoS

- Network shall be QoS enabled, provide end-to-end QoS within the network domain
and support QoS across network domain boundaries
Since emerging high-speed networks are expected to integrate a wide variety of
applications with different traffic characteristics and service requirements, to help with QoS
prioritisation, applications can be classified into four categories with respect to their
requirements: Delay-sensitive and loss-sensitive (e.g. interactive video, gaming), delay-
sensitive but tolerant of moderate losses (e.g. voice), sensitive to loss of data but tolerant
to moderate delays (e.g. interactive data), relatively tolerant to both delay and some limited
loss of information (e.g. file transfer).


- Allow operators to implement policy control for IP multimedia sessions
IP multimedia applications are supported by IP multimedia sessions. Examples of IP
multimedia applications include speech communication, real time multimedia applications,
shared online whiteboards etc. In this environment policy, control determines, for instance,
whether the user has administrative permission to make the reservation.


- Support fixed-mobile convergence, e.g. single solutions for session control,
security, QoS, charging and service provisioning for both fixed and mobile users
Fixed-mobile convergence means alliance of wireline and wireless services. Fixed-mobile
convergence is clearly on the roadmap of operators that want to create additional revenue
streams from new value-added services.


- Latency
Latency is concerned with transport processing in intermediate nodes, with reference to
packet length, TDM size etc. Different types of continuous media require different level of




                                      Page 132 of 168
                             IST IP NOBEL "Next generation
                                                                              Title of the document
                             Optical network for Broadband
                                  European Leadership"                   e1686d81-2994-497c-93f6-
                                                                  b2e913b07f59.docD06_DRAFT_V1
                                                                              6JW2 Andreas zweiter
                                                                       Teil.docD06_DRAFT_V15ejw




latency, bandwidth, delay jitter and they also require guarantees that levels of service can
be maintained.




                                     Page 133 of 168
                  IST IP NOBEL "Next generation
                                                             Title of the document
                  Optical network for Broadband
                       European Leadership"              e1686d81-2994-497c-93f6-
                                                  b2e913b07f59.docD06_DRAFT_V1
                                                              6JW2 Andreas zweiter
                                                       Teil.docD06_DRAFT_V15ejw




8   Conclusions




                         Page 134 of 168
                            IST IP NOBEL "Next generation
                                                                          Title of the document
                            Optical network for Broadband
                                 European Leadership"                 e1686d81-2994-497c-93f6-
                                                               b2e913b07f59.docD06_DRAFT_V1
                                                                           6JW2 Andreas zweiter
                                                                    Teil.docD06_DRAFT_V15ejw




 9      Appendix A: Definition and classification of transport layer
        networks
In the context of this Deliverable, definitions and nomenclature on transport network
reported in the NobelNOBEL Dictionary are adopted.
<<  Editors comment: first draft of the NobelNOBEL Dictionary >>


 9.1    Connectionless and connection oriented layer networks
Most of the existing transport networks are based on a Connection Oriented Circuit
Switched (TDM) (COCS) infrastructure (e.g. SDH). The architecture of this type of
transport network is described in Recommendation G.803.
A Connection Oriented Packet Switched (COPS) infrastructure (e.g. ATM) that uses the
COCS infrastructure may also be provided. This type of network is described in
Recommendation I.326.
A Connectionless Packet Switched (CLPS) infrastructure may be provided either directly
over a COCS infrastructure or over a COPS infrastructure.




                                   Page 135 of 168
                             IST IP NOBEL "Next generation
                                                                             Title of the document
                             Optical network for Broadband
                                  European Leadership"                  e1686d81-2994-497c-93f6-
                                                                 b2e913b07f59.docD06_DRAFT_V1
                                                                             6JW2 Andreas zweiter
                                                                      Teil.docD06_DRAFT_V15ejw




 10     Appendix B: Functional modelling of transport layer
        networks
In the context of this Deliverable, the functional modelling and nomenclature of ITU-T Rec.
G.805 are adopted.
A transport network can be decomposed into a number of independent transport layer
networks with a client/server association between adjacent layer networks. Each layer
network can be separately partitioned in a way that reflects the internal structure of that
layer network or the way that it will be managed.
For further details, see ITU-T Rec. G.805.




                                     Page 136 of 168
                              IST IP NOBEL "Next generation
                                                                                Title of the document
                              Optical network for Broadband
                                   European Leadership"                   e1686d81-2994-497c-93f6-
                                                                   b2e913b07f59.docD06_DRAFT_V1
                                                                               6JW2 Andreas zweiter
                                                                        Teil.docD06_DRAFT_V15ejw




 11      Appendix C: Standardisation Bodies and Fora working on
         architectural requirements and functional models for core
         and metro optical networking

 11.1 ITU-T Project Next Generation Network
The concept of a NGN (Next Generation Network) has been introduced in ITU-T to take
account the new situation in telecommunications, characterised by a lot of factors: open
competition between operators due to the total deregulation of markets, explosion of digital
traffic, e.g. due to the increasing use of internet, increasing demand from users for new
multimedia services, increasing demand from users for a general mobility, etc.
A major goal of the ITU-T Project NGN is to facilitate convergence of networks and
services. In addition, a clear demand from the market for short-term standards in the field
of NGN has been identified, which is leading to standardise a set of Recommendations on
NGN.
The intention of the NGN Project is to coordinate all ITU-T activities related to the
establishment of implementation guidelines and standards for the realisation of a Next
Generation Network. The major task of the Project is to ensure that all elements required
for interoperability and network capabilities to support applications globally across the NGN
are addressed by ITU-T standardization activities.


       11.1.1 Basic characteristics of NGN
In this context, a Next Generation Network (NGN) is defined as a packet-based network
able to provide services including Telecommunication Services and able to make use of
multiple broadband, QoS-enabled transport technologies and in which service-related
functions are independent from underlying transport-related technologies. It offers
unrestricted access by users to different service providers. It supports generalized mobility,
which will allow consistent and ubiquitous provision of services to users.
The NGN is characterized by the following fundamental aspects:

 Packet-based transfer
 Separation of control functions among bearer capabilities, call/session, and application/
  service
 Decoupling of service provision from network, and provision of open interfaces
 Support for a wide range of services, applications and mechanisms based on service
  building blocks (including real time/ streaming/ non-real time services and multi-media)
 Broadband capabilities with end-to-end QoS and transparency
 Interworking with legacy networks via open interfaces
 Generalized mobility




                                      Page 137 of 168
                             IST IP NOBEL "Next generation
                                                                               Title of the document
                             Optical network for Broadband
                                  European Leadership"                    e1686d81-2994-497c-93f6-
                                                                   b2e913b07f59.docD06_DRAFT_V1
                                                                               6JW2 Andreas zweiter
                                                                        Teil.docD06_DRAFT_V15ejw




 Unrestricted access by users to different service providers
 A variety of identification schemes which can be resolved to IP addresses for the
  purposes of routing in IP networks
 Unified service characteristics for the same service as perceived by the user
 Converged services between Fixed/Mobile
 Independence of service-related functions from underlying transport technologies
 Compliant with all Regulatory requirements, for example concerning emergency
  communications and security/privacy, etc.

       11.1.2 NGN capabilities
NGN will have to provide the capabilities (infrastructure, protocols, etc.) to make the
creation, deployment and management of all kinds of services (known or not yet known)
possible. This comprises services using all kinds of media (audio, visual, audiovisual), with
all kinds of encoding schemes and data services, conversational, unicast, multicast and
broadcast, messaging, simple data transfer services, real time and non-real time, delay
sensitive and delay tolerant services. Services with different bandwidth demands from a
few kbit/s to hundreds of Mbit/s, guaranteed or not. Within the NGN, there is an increased
emphasis on service customisation by the Service Providers whereby some of them will
offer their customers the possibility to customise their own services. NGN will comprise
service related APIs (Application Programming Interfaces) in order to support the creation,
provisioning and management of services.
One of the main characteristics of NGN is the decoupling of services and networks,
allowing them to be offered separately and to evolve independently. Therefore, in the NGN
architectures proposed, there is a clear separation between the functions for the services
and the functions for the transport. NGN allows the provisioning of both existing and new
services independently of the network and the access type used.
In NGN the functional entities controlling policy, sessions, media, resources, service
delivery, security, etc, may be distributed over the infrastructure, including both existing
and new networks. When they are physically distributed, they communicate over open
interfaces. Consequently, the identification of reference points is an important aspect of
NGN. New protocols are being standardized to provide the communication between those
functional entities. Interworking between NGN and existing networks such as PSTN, ISDN
and GSM is provided by means of Gateways.
NGN will support both existing and "NGN aware" end terminal devices. Hence terminals
connected to NGN will include analogue telephone sets, fax machines, ISDN sets, cellular
mobile phones, GPRS terminal devices, SIP terminals, Ethernet phones through PCs,
digital set top boxes, cable modems, etc.
Specific issues include the migration of voice services to the NGN infrastructure, Quality of
Service related to real time voice services (bandwidth guarantees, delay guarantees,
packet loss guarantees etc) as well as Security. NGN should provide the security
mechanisms to protect the exchange of sensitive information over its infrastructure, to
protect against the fraudulent use of the services provided by the Service Providers and to
protect its own infrastructure from outside attacks.




                                      Page 138 of 168
                              IST IP NOBEL "Next generation
                                                                                Title of the document
                              Optical network for Broadband
                                   European Leadership"                   e1686d81-2994-497c-93f6-
                                                                   b2e913b07f59.docD06_DRAFT_V1
                                                                               6JW2 Andreas zweiter
                                                                        Teil.docD06_DRAFT_V15ejw




At present, similar services are offered to users both on fixed accesses and on mobile
networks. However, they are still considered up to now as different customers, with
different service configurations and no bridging possible between the different services. A
major feature of NGN will be generalized mobility, which will allow a consistent provision of
services to users, i.e. the users will be regarded as a single person when they use different
access technologies, whatever they are.


       11.1.3 General architectural principles for the NGN
The General architectural principles will provide a basis for NGN similar to that contained in
Recommendation X.200.
The technical objective will be to develop a functional methodology and general model, to
enable an NGN to be described in terms of control functions that can be abstracted and
represented separately from the major areas to be controlled (such as resources, services
and transport).


       11.1.4 Functional architecture methodology model
The Functional architecture methodology model will provide guidelines on how to define a
functional architecture for NGN using a functional approach.
The technical objective will be to develop a functional architecture using the methodology
identified above, to decompose an NGN into an appropriate set of functions. Relationships
and connection between functions will be shown in terms of reference points. Useful
groupings of functions will be described to represent certain practical physical realizations.
Consideration will be given to reference points, which may be candidates, at which
interfaces could be defined.


       11.1.5 End-to-end Quality of Service (QoS)
Work is required to handle both the way in which different end system can reach
agreement on the end-to-end QoS for a call and how the parameters set with this upper
layer protocol can be used to control the lower layer, transport and access level QoS
mechanisms.
For the issue of upper layer QoS control it is felt that a distinction can be made between
telephony, where the work is now almost complete, and the wider topic of QoS for
multimedia which needs work on both a “framework” and the definition of each individual
media stream (video, white board, etc.).
Likewise the control of lower layer QoS mechanisms is best divided into two topics: a
“vertical” protocol linking the upper and lower layer QoS mechanisms (diffserv, etc) and a
lower layer “horizontal” mechanism to link the lower layer QoS control between different
domains and networks.
NGN work on end-to-end QoS should concentrate on:

      Completion of end-to-end QoS class definition for telephony, including voice over
       packet networks




                                      Page 139 of 168
                             IST IP NOBEL "Next generation
                                                                               Title of the document
                             Optical network for Broadband
                                  European Leadership"                    e1686d81-2994-497c-93f6-
                                                                   b2e913b07f59.docD06_DRAFT_V1
                                                                               6JW2 Andreas zweiter
                                                                        Teil.docD06_DRAFT_V15ejw




      Definition of a new end-to-end multimedia QoS class definition framework and a
       method of registering QoS classes of individual media components
      Specification of how to use lower layer QoS mechanism to achieve upper layer QoS
       within the network
      Inter-domain lower layer QoS control
      End user perception of QoS

       11.1.6 Service platforms (APIs)
Two of the key “new” aspects of NGN are the separation of service control and provision
from the underlying network, and the extension of service control for telephony to cover
multimedia.
The required service platforms should offer open interfaces, using APIs (such as PARLAY)
and/or proxy servers, for third party service providers use, the resulting services will need
to be accessible to end users as they roam between networks and, naturally, end-to-end
services should be available between users connected to different networks using different
service providers.
NGN work on service platforms should concentrate on:

      Definition of service control architectures covering both OSA APIs and proxy
       aspects;
      Enhancement of mechanisms to support provision of services across multiple
       networks covering both service roaming and interconnectivity of services;
      Development of mechanisms to support user presence and user control of service
       customisation and profiles;
      Impact of user mobility on service platforms.

       11.1.7 Network management
The emergence of various forms of combined fixed, mobile, IP, access, etc. networks
creates increasing complexities and challenges related to the management of such
networks. This also applies to the management of existing and new services across
different network types.


NGN work on network management should concentrate on:

      Enhancement of the overall “core” network management architecture and definition
       of basic network management services and interfaces to suit NGN requirements
       (fault, performance, customer administration, charging/accounting, traffic and
       routing management)
      Inclusion and application of new architectural concepts and new technologies such
       as TML.




                                      Page 140 of 168
                                  IST IP NOBEL "Next generation
                                                                                 Title of the document
                                  Optical network for Broadband
                                       European Leadership"                  e1686d81-2994-497c-93f6-
                                                                      b2e913b07f59.docD06_DRAFT_V1
                                                                                  6JW2 Andreas zweiter
                                                                           Teil.docD06_DRAFT_V15ejw




           11.1.8 Security
The fact that NGN security is inherent but crucial and is touching many areas and SDOs,
just underlines the strategic importance of this area.
Within NGN, security issues interrelate with architecture, QoS, network management,
mobility, billing and payment.
One of the most significant challenges facing the design of NGN security standards is the
fact that the networks are no longer conceived as a monolithic systems with clear
interfaces. Much of the standardisation work in NGN security has to be based on guides
and principles along with APIs so that a secure network can be built from a given selection
of specific NGN components.
NGN work on security should concentrate on:

          Development of compound security architecture for NGNs. In a further step, this
           NGN security group should devise NGN operational security guidelines.
          Development of NGN specific security protocols and APIs.

           11.1.9 Generalized mobility
Generalized mobility means that users will be regarded as a single person when they use
different access technologies, allowing them to use and manage consistently their services
across existing network boundaries.
At present, the user may be able to roam between similar public wireless accesses, and
nomadism is allowed between some fixed accesses, with strong limitations. In the future,
users will be proposed more and more access technologies and they will require soon to
move between different fixed (e.g. xDSL to cable) and public wireless access of various
technologies (e.g. UMTS to WLAN) and to get access consistently to their set of services.
The general user requirements for mobility should include:

     user ability to change access point and/or terminal, and who benefits from the
      requirements below, can be marked as mobile/nomadic users (see definition in next
      section). This implies that the mobility management functions may be applicable
      only to those users marked as mobile/nomadic;
    user get access from any network access point. This includes all access
      technologies identified above, and the ability to use other networks (see definition of
      roaming below). These possibilities may be limited by subscription;
    user get their services in a consistent manner, depending on the constraints they
      experience in their current situations. This is required for services provided by their
      network operator as well as services provided by a third party;
    user availability and reachability should be known to network functions, and
      possibly to services and applications, including those provided by a third party.
Several capabilities should be considered for mobility: To support

       -     Personal mobility;
       -     Terminal mobility;




                                         Page 141 of 168
                             IST IP NOBEL "Next generation
                                                                              Title of the document
                             Optical network for Broadband
                                  European Leadership"                  e1686d81-2994-497c-93f6-
                                                                 b2e913b07f59.docD06_DRAFT_V1
                                                                             6JW2 Andreas zweiter
                                                                      Teil.docD06_DRAFT_V15ejw




      - A combination of both.
Generalized mobility requires significant evolutions of current network architectures.
Enabling more transparent fixed-wireless broadband communications and mobility across
various access technologies appears as a major issue.
The following requirements for the NGN systems can be derived from the above objectives,
in a mobility management perspective:

           consistent approach from initial 3G systems and fixed systems
           cost reduction (network deployment and operation)
           increased spectrum efficiency
            nomadism, mobility and roaming among different access systems, fixed or
            mobile.
In order to support global mobility in heterogeneous environment, further work is needed to
develop network functions at the control layer:

           Identification and authentication mechanisms
           Access control and authorisation function
           Location management
           Terminal and/or session address allocation and management
           Support of user environment management (VHE)
           User profile management
           Access to user data

       11.1.10 Network control architecture(s) and protocols
Considering the increasingly distributed nature of the control functions in NGN
architectures, there is a need to study Network control reference models encompassing:

        Resource and QoS at the access to the network and in the core network,
        Media processing, transcoding and information transfer
        Call/session control,
        Service control.
The Network control architecture model will take into account the various control related
functional requirements arising from the relevant study areas and will define typical
functional groupings, which interact through reference points.
Examples of functional groupings may include:

          Media access gate (at the network edge), with e.g. firewall, NATP, transfer
           policy enforcement functions …
          Resource control, including e.g. admission control, access request handling, …
          Access session control, including e.g. address allocation, user location, user
           access profile management,
          Service control, including e.g. user registration, user service profile
           management, service requests handling, service interaction management …




                                     Page 142 of 168
                               IST IP NOBEL "Next generation
                                                                               Title of the document
                               Optical network for Broadband
                                    European Leadership"                  e1686d81-2994-497c-93f6-
                                                                   b2e913b07f59.docD06_DRAFT_V1
                                                                               6JW2 Andreas zweiter
                                                                        Teil.docD06_DRAFT_V15ejw




The network control functional models will be used, as the basis to identify reference points
for which there is a need for standardization. This should be based on Recommendation
Y.140. Such reference points will be defined as standard interfaces where the control
protocols will be defined and standardized on the basis of relevant protocol basis, e.g. by
means of profiles for the re-use of already specified protocols, e.g. on the basis of H.248
for Media gate control, of SIP for Call/session control.
The Network control architecture models will take into account of functional requirements at
the network access (user-network interface), at the interfaces between networks (network-
network interfaces) and at the interfaces between networks and service/application
providers (e.g. network-providers interfaces), to the extend the study will determine their
relevance.


       11.1.11 Service capabilities and Service Architecture
Considering the present trends and future evolution of customers requirements for services
involving real time and non-real time, wired and wireless, human-to-human, human-to-
machine and machine-to-machine communications, this study area should:
       -   Address the telecommunication service capabilities that the NGN should
           provide, bearing in mind the separation between applications services and
           networks
       - Develop a suitable service architecture focused on the interfaces that are
           needed to support different business models and seamless communication in
           different environment.
The work should include the backward compatibility with and the evolution from the existing
services and systems.


       11.1.12 Interoperability of Services and Network in NGN
Considering that NGN will involve a large amount of protocols (including various profiles) at
the services and network level, there is a need in the framework of the NGN-2004 project
to ensure the interoperability among systems and networks.

This study area will include in particular:
           -   Specifications for interoperable profiles for complex systems
           -   Specifications for compliance verification of standards
           -   The development of the relevant procedures and documentation, including the
               development of tools.

       11.1.13 Numbering, naming and addressing

                        11.1.13.1 General
Since the NGN would be made up of interconnected heterogeneous networks, using
heterogeneous user access and heterogeneous user devices (fixed, mobile, nomadic, etc)
and that the NGN should provide a “seamless user experience” independent of access




                                       Page 143 of 168
                              IST IP NOBEL "Next generation
                                                                                Title of the document
                              Optical network for Broadband
                                   European Leadership"                   e1686d81-2994-497c-93f6-
                                                                   b2e913b07f59.docD06_DRAFT_V1
                                                                               6JW2 Andreas zweiter
                                                                        Teil.docD06_DRAFT_V15ejw




method and network, work is required on Numbering, Naming and Addressing and so must
be added as an element of the NGN 2004 Project.
Individual users may be identified by name/numbers using a name/number resolution
system that will be able to translate a given name/number into a routable and valid address
in order to establish a transfer (transport).
Examples of such Naming/Numbering schemes may be:

 E.164 numbering scheme
 Unified resource locater (URL) scheme
 Unique name system (e.g. 1800Airways etc.)
 or other naming conventions like developed in the Internet community, like H.323, SIP,
   telephone and mail unified resource identifier (URIs. - Use of International character set
   for URIs is for further study.) etc.
A user who requires access to another user may directly input one of the above-mentioned
identifiers and then either the terminal or the network may translate the user input into an
end-point address, either using internal data or an external database (for example,
accessed via the DNS using enum functionality).

                      11.1.13.2 Achieving a Seamless User Experience
Within the NGN, preference should be given to portable numbering and naming schemes
to provide full control to the end user as well as service providers over his identity.

                      11.1.13.3 Fundamental Principles for Name Resolution
1) Fundamental Requirements
As a public operation network, the NGN shall meet the following requirements for the name
resolution:
   Speed: The name resolution is a necessary process for establishment of a call, and
    speed of the resolution directly impacts the call establishment delay. The call
    establishment delay is strictly specified in real-time communications, and it is directly
    related to the important service index of the call completion ratio.
   Database: A repository of various names, numbers and associated NGN addresses for
    a given user. This database will play a role in facilitating the resolution function and
    could b used to provide additional services like service selection and delivering targeted
    presence information.
   High capacity: As the network is designed to provide national and global services, it
    may have thousands of local control systems and tens of millions of terminal users.
    There are two dimensions to the capacity requirements. First, the database should be
    big enough to contain the name resolution information of the whole network; secondly,
    the resolution system shall be able to handle the numerous concurrent resolution
    requests in the overall network.
   Expandability: As the NGN expands, so should the capacity of the name resolution
    system. The expansion requirements also have two dimensions. First, along with the



                                      Page 144 of 168
                               IST IP NOBEL "Next generation
                                                                                   Title of the document
                               Optical network for Broadband
                                    European Leadership"                    e1686d81-2994-497c-93f6-
                                                                     b2e913b07f59.docD06_DRAFT_V1
                                                                                 6JW2 Andreas zweiter
                                                                          Teil.docD06_DRAFT_V15ejw




    expanding network, the resolution system shall be flexible enough to add new
    information, and expand new resolution levels. Secondly, as new terminal and name
    types are introduced, the resolution system shall be able to expand appropriate name
    resolution information systems. It should be constructed in a way that it could be
    implemented as a distributed name/number resolution system if necessary.
   Reliability: The name resolution system is directly related to the running of the NGN, so
    it should have carrier class reliability. It shall have two capabilities in the architecture.
    First, it shouldn't be a single-point of failure. Secondly, it should have excellent load
    balancing mechanism. Good configuration and arrangement shall be conducted to
    meet the capacity requirements during the network planning.
   Security: The name resolution data are important network data that may directly impact
    the operation of the network, and they are sensitive commercial data reflecting the
    structure and policy of the network operations. Accordingly, the name resolution system
    shall be a special system used only by this network, and certain security measures
    shall be in place. The security is mainly maintained by the means of user access
    authentication, data security, network data synchronization and fault recovery.

2) Basic techniques
As mentioned above, in the NGN call and connection are completely separate from each
other and the call/session control is an end-to-end process only involving the call control
entities at the two ends. Therefore, an independent name resolution system is necessary.

                       11.1.13.4 Routing Protocols
Considering, that multiple routing protocols are defined and used in the markets, the NGN
project should investigate the implications of the most promising. Whilst currently IPv4 is
the most widely used and understood protocol, the next version, IPv6, provides significant
enhancements in functionality but also significant challenges for migration towards future
wireline and wireless network environments. It is widely expected that the IPv6 protocol will
be a useful carrier for transport, control, and management flows of the NGN. In alignment
with the reference model and functional architecture of NGN that are currently developed
by the JRG-NGN, other protocols relevant to IPv6 and mechanisms using it should be
investigated.
New work items on IPv6 over NGN are as follows:

       The IPv6 based protocols and mechanisms for the U-plane, C-plane and M-plane
        flows in alignment with the NGN functional architecture and reference model (e.g.,
        IPv6 address auto configuration, DNS service, service discovery, etc.)
       The integrated wireline and wireless network architecture in consideration for the
        IPv6 capability within the scope of NGN
       The media gateway architectures and functional blocks to use the IPv6 protocol as
        well as the IPv4 protocol (The interworking or conversion functions between the
        IPv4 and the IPv6 may be included.)
       The transport network architecture using the IPv6-based control and management
        flows




                                       Page 145 of 168
                        IST IP NOBEL "Next generation
                                                                        Title of the document
                        Optical network for Broadband
                             European Leadership"                 e1686d81-2994-497c-93f6-
                                                           b2e913b07f59.docD06_DRAFT_V1
                                                                       6JW2 Andreas zweiter
                                                                Teil.docD06_DRAFT_V15ejw




                 11.1.13.5 Related Standards
   Y.NGN-Overview General overview of NGN functions and characteristics
   Y.GRM-NGN General reference model for NGN
   Y.NGN-FRM Functional architecture model (ex-Functional requirements and
    architecture of the NGN)
   Y.NGN-SRQ NGN service requirements
   Y.NGN-MOB Mobility management requirements and architecture for NGN
   Y.NGN-MAN Framework for manageable IP network
   Y.NGN-MIG Migration of networks (including TDM networks) to NGN
   Y.NGN-CON Regulatory consideration of the NGN
   Y.e2eqos End-to-end QoS architecture for IP networks evolving into NGN
   Y.123.qos A QoS architecture for Ethernet-based IP access network
   Y.NGN-TERM Next Generation Networks terminology: Terms and definitions




                                Page 146 of 168
                                    IST IP NOBEL "Next generation
                                                                                                  Title of the document
                                    Optical network for Broadband
                                         European Leadership"                             e1686d81-2994-497c-93f6-
                                                                                   b2e913b07f59.docD06_DRAFT_V1
                                                                                               6JW2 Andreas zweiter
                                                                                        Teil.docD06_DRAFT_V15ejw




  11.2 ITU-T SG13 (2001-2004)
ITU-T SG13 "Multi-protocol and IP-based networks and their internetworking" is
responsible for studies relating to:
        Inter-networking of heterogeneous networks encompassing multiple domains,
         multiple protocols and innovative technologies with a goal to deliver high-quality,
         reliable networking. Specific aspects are architecture, interworking and adaptation,
         end-to-end considerations, routing and requirements for transport.
Questions where Standards, in the NobelNOBEL areas of studies, are developed
WP 1/13 - Project management and Coordination
Question         Title
12/13            Global Coordination of Network Aspects
15/13            General Network Terminology including IP Aspects
WP 2/13 - Architectures and Internetworking Principles
Question         Title
1/13             Principles, Requirements, Frameworks and Architectures for an Overall Heterogeneous Network
                 Environment
5/13             Network Interworking including IP Multiservice Networks
10/13            Core Network Architecture and Interworking Principles
16/13            Telecommunication Architecture for an Evolving Environment
WP 3/13 - Multi-protocol Networks and Mechanisms
Question         Title
3/13             OAM and Network Management in IP-Based and Other Networks
11/13            Mechanisms to Allow IP-Based Services Using MPLS to Operate in Public Networks
WP 4/13 - Network Performance and Resource Management
Question        Title
4/13            Broadband and IP Related Resource Management
6/13            Performance of IP-Based Networks and The Emerging Global Information Infrastructure
8/13            Transmission Error and Availability Performance
9/13            Call Processing Performance




  11.3 ITU-T SG15 (2001-2004)
ITU-T SG15 "Optical and other transport networks" is the focal point in ITU-T for studies on
optical and other transport networks, systems and equipment.
This encompasses the development of transmission layer related standards for the access,
metropolitan and long haul sections of communication networks.
Specifically regarding optical technology ITU-T SG15 work is intended to track
telecommunications standardization activities of optical transport networks and
technologies (OTNT) both within the ITU-T and across other organizations. The primary
goals are to facilitate better cooperation between organizations and more coordination of




                                               Page 147 of 168
                                    IST IP NOBEL "Next generation
                                                                                                  Title of the document
                                    Optical network for Broadband
                                         European Leadership"                                 e1686d81-2994-497c-93f6-
                                                                                       b2e913b07f59.docD06_DRAFT_V1
                                                                                                   6JW2 Andreas zweiter
                                                                                            Teil.docD06_DRAFT_V15ejw




activities. The desired result is the development of more interoperable and widely accepted
standards concerning optical functions, namely:
       Multiplexing function
       Cross connect function, including grooming and configuration
       Management functions
       Physical media functions.

Questions where Standards, in the NobelNOBEL areas of studies, are developed

  Question                                                        Title

WP 3/15 – OTN Structure

    9/15       Transport equipment and network protection/restoration

    10/15      ATM and Internet Protocol (IP) equipment

    11/15      Signal structures, interfaces and interworking for transport networks

    12/15      Technology Specific Transport Network Architectures

    14/15      Network management for transport systems and equipment

WP 4/15 – OTN Technology

    15/15      Characteristics and test methods of optical fibres and cables

    16/15      Characteristics of optical systems for terrestrial transport networks

    17/15      Characteristics of optical components and subsystems

    18/15      Characteristics of optical fibre submarine cable systems




                                               Page 148 of 168
                             IST IP NOBEL "Next generation
                                                                              Title of the document
                             Optical network for Broadband
                                  European Leadership"                  e1686d81-2994-497c-93f6-
                                                                 b2e913b07f59.docD06_DRAFT_V1
                                                                             6JW2 Andreas zweiter
                                                                      Teil.docD06_DRAFT_V15ejw




 11.4 OIF
"The mission of the Optical Internetworking Forum (OIF) is to foster the development and
deployment of interoperable products and services for data switching and routing using
optical networking technologies. The OIF will encourage co-operation among telecom
industry participants including equipment manufacturers, telecom service providers and
end users; promote global development of optical internetworking products; promote
nationwide and worldwide compatibility and interoperability; encourage input to appropriate
national and international standards bodies; and identify, select, and augment as
appropriate and publish optical internetworking specifications drawn from national and
international standards.

Being the only industry group uniting representatives from the data and optical networks,
OIF's purpose is to accelerate the deployment of interoperable, cost-effective and robust
optical internetworks and their associated technologies. Optical internetworks are data
networks composed of routers and data switches interconnected by optical networking
elements."

Working Groups of the Technical Committee focus on specific areas where Implementation
Agreements are developed.

       Architecture and Signalling Working Group

       Carrier Working Group

       Interoperability Working Group

       OAM&P Working Group

       Physical and Link Layer Working Group

       Physical Layer User Group

 11.5 IETF
The Internet Engineering Task Force (IETF) is a large open international community of
network designers, operators, vendors, and researchers concerned with the evolution of
the Internet architecture and the smooth operation of the Internet. It is open to any
interested individual.
The actual technical work of the IETF is done in its working groups, which are organized by
topic into several areas:
      Applications Area
      General Area
      Internet Area
      Operations and Management Area
      Routing Area



                                     Page 149 of 168
                       IST IP NOBEL "Next generation
                                                                  Title of the document
                       Optical network for Broadband
                            European Leadership"              e1686d81-2994-497c-93f6-
                                                       b2e913b07f59.docD06_DRAFT_V1
                                                                   6JW2 Andreas zweiter
                                                            Teil.docD06_DRAFT_V15ejw




     Security Area
     Sub-IP Area
     Transport Area
                                                                                      Formatted: Bullets and Numbering
1




                              Page 150 of 168
                             IST IP NOBEL "Next generation
                                                                               Title of the document
                             Optical network for Broadband
                                  European Leadership"                    e1686d81-2994-497c-93f6-
                                                                   b2e913b07f59.docD06_DRAFT_V1
                                                                               6JW2 Andreas zweiter
                                                                        Teil.docD06_DRAFT_V15ejw




 12     Appendix D: Transparent Optical Network and Wavelength
        ServicesPurpose, Scope and document overview
Different people working in the NobelNOBEL project have different areas of expertise, even
people working within the same workpackagework package as for example in WP5. It has
come to the author's attention there are confusions when people talk about transparent
optical networks and wavelength services. Hence this short document tries to explain very
concisely what is meant about transparency, dynamism, switched network, wavelength
service and switched service, with the background of transparent optical networks.
Therefore this document will help experts on the physical layers of optical networks
understanding network technologies and service aspects in which optical network are or
may be involved.
It is the intention of the authors to keep this document short and clear, and to explain the
terms at a conceptual level only.
On the other hand this document tries also to engage people working in other
workpackageswork packages on a constructive debate about transparency and its impact
on the services a Network Service Provider is offering. A further goal is to help defining a
wavelength service scenario to be studied within other workpackageswork packages.
At the beginning of this document in Section 12.2 we present the terminology we want to
clarify in this document. The terms embrace the physical layer and the service layer where
sometimes there is confusion in resolving their meaning because the similarity of the words
used.
Section 12.3 defines the meaning of transparency we are using, and Section 12.4 what is
meant by static and dynamic network. Section 12.5 explains what is understood by
switched network. The next two sections move to the service level explaining what a
wavelength service is in Section 12.6 either using just a transparent optical network or a
switched transparent optical network. Finally, Section 12.7 defines a switched wavelength
service.


 12.2 Introduction
The terms we want to clarify are:
   1. Transparent Optical Network
   2. Static & Dynamic Transparent Optical Networks
   3. Switched Transparent Optical Network
   4. Wavelength Service
   5. Switched Wavelength Service
Transparent Optical Network and its Dynamism are terms that have already been
discussed in detail in [Andrew], and here we will only briefly present. These definitions and
the Switched Transparent Optical Network refer only to the technology used in the network



                                      Page 151 of 168
                              IST IP NOBEL "Next generation
                                                                                Title of the document
                              Optical network for Broadband
                                   European Leadership"                   e1686d81-2994-497c-93f6-
                                                                   b2e913b07f59.docD06_DRAFT_V1
                                                                               6JW2 Andreas zweiter
                                                                        Teil.docD06_DRAFT_V15ejw




whilst the last two refer to network services. There is a relation between these two sets of
concepts in the sense that the latter network service is impossible without certain network
technology.


 12.3 Transparent Optical Network
Previous to defining what we mean by Static and Dynamic Transparent Optical Network we
state the definition of Transparent Optical Network that WP5 agreed and presented in
Section 2.1 of deliverable [Andrew]. A Transparent Optical Network is an infrastructure
where the transmission of the optical signal is independent of the specific characteristics of
the actual data to be transported through the optical layer.
Ideally, an all-optical network (i.e. without o/e/o conversions) is transparent, although the
size and reach will be limited by signal degrading effects (e.g. OSNR). It is one of the major
objectives of WP5 to investigate what these limits are and try defining engineering rules to
help designing and planning transparent optical networks.
Keeping the traffic signals in the optical domain could help in reducing network costs
(CapEx) by eliminating o/e/o conversions within the network (conversions only at the
ingress and egress points), and indirectly operational costs (OpEx), as network equipment
is reduced.
Transparency is not a service, but implies a technology intended to benefit the network
operator or Network Service Provider (NSP). Potential advantages for NSPs are:
      CapEx savings due to savings in o/e/o conversions
      Possibly OpEx savings due to reduced equipment


 12.4 Static & Dynamic Transparent Optical Networks
We are interested in the infrastructure of a transparent optical network where network
entities such as wavelengths, wavelength multiplex, optical fibres, and cross-connecting
subnetworks support end-to-end connections in the optical network (i.e. optical channels or
light-paths). The static or dynamic nature of this network refers to the operation time-scales
of this infrastructure.
A Static Transparent Optical Network (see section 2.5 in [Andrew]) is a set of nodes
connected by optical transmission links in which events happen at such slow rate
(connections are labelledlabeled permanent) that planning and provision of connections
can be accomplished successfully through the Management System, even when it involves
manual operation. Moreover, the network is engineered such that even in case of failures
the network remains stable.
Hence, in a static transparent optical network the network operator needs engineering rules
to assess the maximum path length, maximum number of crossed nodes along the path,
and maximum number of wavelengths any link along the path can have. In order to
maintain the connections, means for traffic performance monitoring are necessary to
support Performance Management and Fault Management. Further it is necessary to
provide connection traceability for Configuration Management, Security Management, and
Accounting Management.




                                      Page 152 of 168
                                  IST IP NOBEL "Next generation
                                                                                      Title of the document
                                  Optical network for Broadband
                                       European Leadership"                       e1686d81-2994-497c-93f6-
                                                                           b2e913b07f59.docD06_DRAFT_V1
                                                                                       6JW2 Andreas zweiter
                                                                                Teil.docD06_DRAFT_V15ejw




In a static transparent optical network, path restoration at the optical layer is not possible13,
and therefore resilience must be implemented using protection mechanisms. This means
that spare capacity must be provisioned and reserved to protect working links and/or paths
(dedicated or shared).
A Dynamic Transparent Optical Network is an infrastructure in which its configuration
changes in time-scales for which planning and manual intervention is not possible. Such a
network requires dynamic/agile network elements, which enable the rapid configuration of
the infrastructure. Operation times are such that they cannot be left to the Management
Plane alone.
An optical network is a set of switching nodes linked by fibre cables. Network nodes are
sites where traffic is generated, terminated and/or carried through. Generated traffic is
converted from electrical to optical format before launching it into the optical network.
When traffic is terminated it is converted back into electrical format. What happens with the
through traffic? In a transparent network (no o/e/o conversions along the path but at the
end systems) we can route it by 'jumpering' using patch panels (analogous to PDH
networks, see Figure 42 Error! Reference source not found.), or it could be routed via an
Optical Cross-connect (OXC) or Optical Add/Drop Multiplexer (OADM). Patch panels can
only be used in static networks needing manual configuration thus being prone to errors.
They also lead to complex fibre management and are very inflexible too. OXCs and/or
OADMs bring flexibility to the network, which can be reconfigured either 'manually' using a
Management System in a static network or automatically using Control Plane technology in
a dynamic network for which are essential components, see Figure 43 (Error! Reference
source not found.). The degree of network flexibility will be given by the reconfiguration
(and maybe tuning) capability of the intermediate OXC/OADM along the optical path.
Different possibilities of node architecture, capability, and speed (among others) regarding
reconfigurability are presented and discussed in Section 3.2.3 of [Andrew].




13
     Restoration could be implemented at electronic levels like e.g. OTH, SDH, etc.




                                           Page 153 of 168
                           IST IP NOBEL "Next generation
                                                                          Title of the document
                           Optical network for Broadband
                                European Leadership"                 e1686d81-2994-497c-93f6-
                                                              b2e913b07f59.docD06_DRAFT_V1
                                                                          6JW2 Andreas zweiter
                                                                   Teil.docD06_DRAFT_V15ejw




Figure 414242: Illustration of a node with patch panel in a static transparent optical
                                       network




  Figure 424343: Illustration of a node in a dynamic transparent optical network



                                  Page 154 of 168
                               IST IP NOBEL "Next generation
                                                                                   Title of the document
                               Optical network for Broadband
                                    European Leadership"                     e1686d81-2994-497c-93f6-
                                                                      b2e913b07f59.docD06_DRAFT_V1
                                                                                  6JW2 Andreas zweiter
                                                                           Teil.docD06_DRAFT_V15ejw




 12.5 Switched Transparent Optical Network
A Switched Optical Network is a dynamic network with Control Plane technology [McGuire].
Such network is termed ASON (Automatic Switched Optical Network), and its architecture
has been standardised by ITU-T [ASON]. This type of network is also briefly discussed in
Section 2.5 of [Andrew].
The two major functions that can be implemented using Control Plane technology are
routing and signalling. The latter relies on a routing algorithm and has two main roles:
           a) set up, maintenance and tear down of paths across the network
           b) provision of communications between two end systems requesting the
              connection.
To realise a) it is necessary to define Network-Network Interfaces (NNI) that enable
communications between intermediate switching nodes along the network path. To realise
b) it is necessary to define User-Network Interfaces (UNI) that enable communications
between end systems. UNIs and NNIs are able to support real-time connection requests.
Note that whilst realising b) needs the realisation of a) as well, the contrary is not true, and
it is possible to automate path set up, maintenance, and tear down between the
intermediate node switches, on request by the network management and not through an
UNI (Figure 43Figure 44Figure 44). This type of connection is opaque to the user and will
benefit the NSP in speeding up provision. There are two types of NNI namely the Internal-
NNI (I-NNI) and the External-NNI (E-NNI). The former is used between subnetworks within
an administrative (or routing) domain whilst the latter is used between administrative (or
routing) domains14.




14
 The difference is that the E-NNI implements Call Control as well as Resource Discovery,
Connection Control, Connection Selection, and Connection Routing [3]




                                       Page 155 of 168
                              IST IP NOBEL "Next generation
                                                                                 Title of the document
                              Optical network for Broadband
                                   European Leadership"                   e1686d81-2994-497c-93f6-
                                                                   b2e913b07f59.docD06_DRAFT_V1
                                                                               6JW2 Andreas zweiter
                                                                        Teil.docD06_DRAFT_V15ejw




                                                       Network Management requests
                                                       connection



         ES: End System

                                                                            ES
         ES
                              Transparent Optical Layer
       Permanent                NNI                  NNI                  Permanent
       connection                                                         connection

                              Signalling           Signalling
                              messages             messages




     Figure 434444: Connection configuration requested by Mgmt. System: Soft-
                             Permanent Connection
Control Plane technology also offers the possibility of service restoration. When using
protection, resources are provisioned and allocated in advance and therefore, after a
failure occurs, service can be recovered with minimal disruption by switching the traffic
over to the protection resources. When using restoration there are no provisioned
resources allocated to the service and a routing algorithm will look for available resources
across the network to switch traffic over after a failure has occurred. Restoration is used to
recover those services that do not use protection mechanisms or to recover those using
protection mechanisms from dual network failure. Restoration relies on available spare
capacity in the network whilst protection relies on redundant capacity being assigned.
In opaque optical networks soft-permanent connections do not add management functions
to the static case but transfers much of the configuration management to the control plane
(automatic processes). Because provision times will be down to seconds, this is also the
expected lapse between set up and/or tear down of different connections. Therefore the
inherent dynamism when using control plane technology may impact the network
performance especially in transparent optical networks, which has to be studied. Special
precaution must be taken when using restoration as the new available path could be
beyond the physical performance limits and the time scales involved are of the order of
seconds. Therefore dynamism and restoration in transparent optical networks poses very
serious problems for routing algorithms and network design. For more detailed information
see [Andrew].




                                      Page 156 of 168
                                IST IP NOBEL "Next generation
                                                                                   Title of the document
                                Optical network for Broadband
                                     European Leadership"                     e1686d81-2994-497c-93f6-
                                                                       b2e913b07f59.docD06_DRAFT_V1
                                                                                   6JW2 Andreas zweiter
                                                                            Teil.docD06_DRAFT_V15ejw




               ES: End System

                                                                                  ES
           ES
                                      Transparent Optical Layer
                                                                            UNI
                    UNI
                                    NNI                  NNI
       Signalling
       messages                                                               Signalling
                                                                              messages
                                  Signalling           Signalling
                                  messages             messages




  Figure 444545: Connection configuration requested by ES: Switched Connection
Switched networks offer the opportunity for new network services. However, these depend
on the specific control plane technology implemented and on the provider's ability to design
the network and the service. An illustration of a switched connection between end systems
through UNIs and NNIs is shown in Figure 44Figure 45Figure 45. These connections need
the development of new management/control processes such as address resolution,
authentication, etc. It is important to note that it is necessary to deploy transponders (i.e.
electronics) at an UNI reference point, and therefore this scenario is not the prime concern
within WP5.
A transparent optical network along with a suitable Control Plane opens the possibility of
further cost savings (apart from new services) on top of those offered by a static
transparent optical network mainly thanks to simplified provisioning and reduced
provisioning times.
This type of network needs the development of new management techniques. If these
capabilities were successfully brought about, then the automatic switching capability could
reduce OpEx and maybe CapEx, as the installed capacity in the network should be used
more efficiently.
Note that this is not a service either, but also for the benefit of the network operator. What
is true, however, is that this could enable automated services. It would be a sort of
Transparent ASON, maybe ASTON (Automatic Switched Transparent Optical Network).
Summarising, Switched Transparent Optical Networks
           -    Reduce provisioning times
           -    Could reduce OpEx by automating provisioning
           -    Maybe reduce CapEx by using network capacity more efficiently
           -    Makes transparent network dynamic and thus impacts design and
                management
           -    Enables new network services
           -    Allows the capability of restoring unprotected paths



                                          Page 157 of 168
                              IST IP NOBEL "Next generation
                                                                                  Title of the document
                              Optical network for Broadband
                                   European Leadership"                     e1686d81-2994-497c-93f6-
                                                                     b2e913b07f59.docD06_DRAFT_V1
                                                                                 6JW2 Andreas zweiter
                                                                          Teil.docD06_DRAFT_V15ejw




 12.6 Wavelength Service
Customers don't care what providers have in their networks. They care about cost and
service capabilities (time to get service activated, availability, etc). A wavelength service is
the offer of a specific wavelength connection between 2 end systems, that is, instead of the
connection being supported by a wavelength (usually the client would be oblivious of the
wavelength used) the connection becomes the wavelength itself (at least at the
ingress/egress points) and therefore it has to be specified within a certain tolerance. The bit
rate has to be specified too, sometimes as a maximum, but sometimes offered at a specific
bit rate only. One aspect of this service is that it can be offered as a managed or
unmanaged service. In a managed service, the client traffic must be handed over using an
agreed format and bit rate. The NSP handles the overhead to ensure connection integrity
and proper connection configuration. At the destination end system the NSP hands it back
to the client. In an unmanaged service, the NSP transfers the user traffic transparently
across its network to the destination end system. There is no performance monitoring and
no ability to localise faults within the NSP's network. In this latter case the bit rate can be
specified just as a maximum allowed value.
Another important point regarding network services is that as we descend through the
layers of the transport network, the number of connections decreases thus there is less
statistical gain to be realised. Moreover, the holding times increase dramatically and churn
rates decrease (the duct being the slowest).
Currently, it is impossible to offer a truly fully managed wavelength service as this means
you would need the means to manage an analogue signal (wavelengths are analogue
signals) and offer guarantees of their quality. This case is analogous to what nowadays is
called the Optical Transport Network, where the carrier can switch wavelengths but the
switching nodes are electrical. This means that the only way of offering a managed
wavelength service is using G.872 or Optical Transport Network (OTN) and G.709 or
Optical Transport Hierarchy (OTH), where the carrier has access to digital information for
management purposes.
What happens when the wavelength service uses a transparent optical network? Here a
set of possibilities arises presented in the next sections.


       12.6.1 Wavelength Service over a Transparent Optical
             Network
The simplest scenario is to consider the wavelength as "nailed up" across the transparent
optical network (static network). In this case the transparent path can be engineered such
that the signal quality is guaranteed at the destination end system, and there is no need for
complex management. This method has the drawback of long lead times and high cost,
although the cost is expected to be lower than current wavelength services through non-
transparent optical networks (cost reduction due to no o/e/o conversions only).




                                       Page 158 of 168
                            IST IP NOBEL "Next generation
                                                                            Title of the document
                            Optical network for Broadband
                                 European Leadership"                  e1686d81-2994-497c-93f6-
                                                                b2e913b07f59.docD06_DRAFT_V1
                                                                            6JW2 Andreas zweiter
                                                                     Teil.docD06_DRAFT_V15ejw




    Figure 454646: Wavelength service over a static transparent optical network



       12.6.2 Wavelength Service over a Switched Transparent
             Optical Network
In order to further reduce costs, it could be possible to think of deploying an automatic
switched transparent optical network. This would allow bringing lead times down to the
order of minutes or even seconds. To offer this kind of service, it is mandatory to have
management capabilities in the transparent optical network to ensure the integrity of the
signal at the destination end system. In this scenario network dynamism is likely to
increase the problems in designing and engineering the network.
This service can be delivered using soft-permanent connections as shown in Figure
46Figure 47Figure 47. The route is prompted by the Management System and established
using the Optical Connection Controllers (OCC) that communicatecommunicates through
NNI interfaces. The intermediate nodes are automatically configured using the Control
Plane.




                                    Page 159 of 168
                             IST IP NOBEL "Next generation
                                                                              Title of the document
                             Optical network for Broadband
                                  European Leadership"                   e1686d81-2994-497c-93f6-
                                                                  b2e913b07f59.docD06_DRAFT_V1
                                                                              6JW2 Andreas zweiter
                                                                       Teil.docD06_DRAFT_V15ejw




  Figure 464747: Wavelength Service over a switched transparent optical network
                        using soft-permanent connection


 12.7 Switched Wavelength Service over a Switched Transparent
      Optical Network
Switched services are those where connections are established by dialling-up. PSTN is an
example of switched service. When one or more characteristics of the connection such as
bandwidth, protection, etc., are negotiated at the request stage the service is known as
bandwidth on demand. Therefore a switched wavelength service is what could be called as
wavelength on demand. A major difficulty would be to plan and dimension the network
such that the Quality of Service reaches a certain target, as the number of customers
would be very small and holding times very diverse (unknown statistics).
Note that it would be impossible to offer a wavelength switched service in a non-automatic
optical network. Figure 47Figure 48Figure 48 illustrates this service where a customer dials
up a wavelength connection between two end systems when they require the connection
and for the time they need it.




                                     Page 160 of 168
                            IST IP NOBEL "Next generation
                                                                                  Title of the document
                            Optical network for Broadband
                                 European Leadership"                        e1686d81-2994-497c-93f6-
                                                                      b2e913b07f59.docD06_DRAFT_V1
                                                                                  6JW2 Andreas zweiter
                                                                           Teil.docD06_DRAFT_V15ejw




             Figure 474848: Illustration of a switched wavelength service
This other trusted administrative domain can be e.g. a different NSP or can be another
transparent network belonging to the same NSP. This case would cover the wavelength
service scenario across a 'translucent' network as defined in [Andrew]. We think this is a
possible longer-term service scenario.




                                Transparent Optical Network
                                                                              Network Z
        Network A
                                      I-NNI           I-NNI

                    E-NNI                                             E-NNI

                              I-NNI                           I-NNI
                                              I-NNI




Figure 484949: Illustration of a switched wavelength service through E-NNI (NNIs are
interfaces between the nodes' OCCs)




                                      Page 161 of 168
             IST IP NOBEL "Next generation
                                                        Title of the document
             Optical network for Broadband
                  European Leadership"              e1686d81-2994-497c-93f6-
                                             b2e913b07f59.docD06_DRAFT_V1
                                                         6JW2 Andreas zweiter
                                                  Teil.docD06_DRAFT_V15ejw




References




                    Page 162 of 168
                              IST IP NOBEL "Next generation
                                                                                  Title of the document
                              Optical network for Broadband
                                   European Leadership"                    e1686d81-2994-497c-93f6-
                                                                    b2e913b07f59.docD06_DRAFT_V1
                                                                                6JW2 Andreas zweiter
                                                                         Teil.docD06_DRAFT_V15ejw




 13     Appendix E: Applications questionnaire

 13.1 Introduction
In the questionnaire the following parameters should be considered for the set of emerging
application described in chapter 3.
The considered parameters were:
      Unavailability: the probability that a customer tries to use the application and this
       application is not available at the moment (for every reason)
      Latency: all the issues related to latency: the time to exchange data between the
       end-points and the difference of latency.
      possibility of changing connection parameters: marking by 5 it does mean that
       the possibility of changing the connection parameters (e.g. QoS, connectivity or bit-
       rate) is very important for the success of the application, instead marking by 1 does
       mean that the possibility of changing the connection param.eter is not necessary.
      breaks during application in use: it refers to the breaks due on some network
       impairments that the final user can tolerate. The breaks may last some
       seconds/minutes of abort at all the application. Marking by 5 does mean the user
       can not tolerate any break; instead marking by 1 does mean that can tolerate some
       breaks during the use of the application.
      set-up / tear-down time: it refers to the application. The time elapsing between the
       requiring by the customer to use the application and the real use (set-up) or the
       time between the finish of use the application and the possibility of requiring again
       for the same application.
      Price: all the issues related to price as it appears to the final user (e.g. installation
       fee, price policies, …). Marking by 5 this item it does mean a low price is very
       important for the success of the application; instead marking by 1 does mean that
       the level of price is not important for the success of the application.
The outcomes extracted from the applications questionnaire might be interesting to
understand:
      which characteristics shall have the new applications to be competitive with the
       incumbent ones;
      which characteristics shall have new applications that don’t have any incumbent
       application that deal in the same field;
      which parameters are the most important that we have to take in account;
      which applications might be the most profitable from the provider/operator point of            Formatted: Bullets and Numbering
       view.
It is important to note that there are several performance parameters and then there is the
price (economical parameter).



                                      Page 163 of 168
                                IST IP NOBEL "Next generation
                                                                                  Title of the document
                                Optical network for Broadband
                                     European Leadership"                   e1686d81-2994-497c-93f6-
                                                                     b2e913b07f59.docD06_DRAFT_V1
                                                                                 6JW2 Andreas zweiter
                                                                          Teil.docD06_DRAFT_V15ejw




In particular, for each of these parameters, a valuation was given by a number between 1
and 5 where “1” means no strong requirements and “5” very strong requirements. About
price parameter “1” means that the provider can not care about a policy geared towards to
hold low prices, “5” means that a low price is essential for the success of the application.
The goal of this questionnaire is to understand an applications might be successful both for new
applications that try to replace an incumbent one (e.g. VoIP vsvs. traditional PSTN) and new
applications.
We try to concentrate the study from the customer point of view.
                       1 = not important      … 5 = very important




                                         Page 164 of 168
                                  IST IP NOBEL "Next generation
                                                                                                                                  Title of the document
                                  Optical network for Broadband
                                       European Leadership"                                           e1686d81-2994-497c-93f6-
                                                                                               b2e913b07f59.docD06_DRAFT_V1
                                                                                                           6JW2 Andreas zweiter
                                                                                                    Teil.docD06_DRAFT_V15ejw




                                                                                          application in use


                                                                                                               set-up/tear down
                                                                                          breaks during
                                              unavailability




                                                                         possibility of


                                                                         parameters
                                                                         connection
                                                                         changing
                                              very low



                                                               latency




                                                                                                                                      price
   Storage

    - Back-Up / Restore

    - Storage on demand (SoD)

    - Asyncrhonous mirroring

    - Synchronous mirroring
   Grid computing

    - Compute Grid

    - Data Grid

    - Utility Grid
   Multimedia
    - Video on Demand (entertainment
   quality, similar to DVD)
    - Video Broadcast (IP-TV),
   entertainmant quality similaar to

    - Video Download
    - Video Chat (SIF quality, no real-
   time coding penalty)

    - Narrowband Voice, data (VoIP,...)

    - Telemedicine (disgnostic)

    - Gaming

    - Digital distribution, digital cinema
    - Video conference (PAL broadcast
   quality 2.0 realtime coding penalty)

                           Table 333333: Questionnaire form sheet




 13.2 Results
The questionnaire was answered by 70 participants during the NobelNOBEL plenary
meeting in Stockholm, in September 2004.
The following figures are showing a summary of the results.



                                             Page 165 of 168
                                           IST IP NOBEL "Next generation
                                                                                                                       Title of the document
                                           Optical network for Broadband
                                                European Leadership"                                             e1686d81-2994-497c-93f6-
                                                                                                          b2e913b07f59.docD06_DRAFT_V1
                                                                                                                      6JW2 Andreas zweiter
                                                                                                               Teil.docD06_DRAFT_V15ejw




Figure 49Figure 50Figure 50 shows which are the most important parameters to take in
account so that an application might have market success. itIt is possible to note that the
focus is addressed in two main directions: reliability (first and forth columns) and price (last
column).

  4


 3,5


  3


 2,5


  2


 1,5


  1


 0,5


  0
       very low unavailability   latency      possibility of changing   breaks during application   set-up/tear down    price
                                              connection parameters              in use




                                  Figure 495050: importance of parameters
Figure 50Figure 51Figure 51 depicts the requirements for each considered application. The
histograms show the average valuation about performance parameters (histogram) and
price (line). The outcome of this analysis is that for multimedia applications (except
telemedicine) a low price is very important for success and for applications that have a
traditional incumbent (voice, TV) a good level of performance is required.




                                                      Page 166 of 168
                                                                                                                         IST IP NOBEL "Next generation
                                                                                                                                                                                                                                                                                                                                                                                                                     Title of the document
                                                                                                                         Optical network for Broadband
                                                                                                                              European Leadership"                                                                                                                                                                                                 e1686d81-2994-497c-93f6-
                                                                                                                                                                                                                                                                                                                                            b2e913b07f59.docD06_DRAFT_V1
                                                                                                                                                                                                                                                                                                                                                        6JW2 Andreas zweiter
                                                                                                                                                                                                                                                                                                                                                 Teil.docD06_DRAFT_V15ejw




  5                                                                                                                                                                                                                                                                                                         Multimedia
 4,5                                                                                                                    Grid computing
  4
                            Storage
 3,5
  3
 2,5
  2
 1,5
  1
 0,5
  0
                                                                                                                                                     Utility Grid




                                                                                                                                                                                                                                                                            Video Download




                                                                                                                                                                                                                                                                                                                                                                                                                                                             Video conference (PAL broadcast quality 2.0
                                                                               Synchronous mirroring
                                                      Asyncrhonous mirroring




                                                                                                                                        Data Grid




                                                                                                                                                                                                                                                                                             Video Chat (SIF quality, no real-time coding



                                                                                                                                                                                                                                                                                                                                            Narrowband Voice, data (VoIP,...)
        Back-Up / Restore




                                                                                                                                                                                                                           Video Broadcast (IP-TV), entertainmant quality




                                                                                                                                                                                                                                                                                                                                                                                                            Gaming
                                                                                                       Grid computing




                                                                                                                                                                    Multimedia


                                                                                                                                                                                 Video on Demand (entertainment quality,
                                                                                                                         Compute Grid
                            Storage on demand (SoD)




                                                                                                                                                                                                                                                                                                                                                                                                                      Digital distribution, digital cinema
                                                                                                                                                                                                                                                                                                                                                                                Telemedicine (disgnostic)




                                                                                                                                                                                                                                                                                                                                                                                                                                                                       realtime coding penalty)
                                                                                                                                                                                            similar to DVD)


                                                                                                                                                                                                                                         similaar to DVD




                                                                                                                                                                                                                                                                                                              penalty)




       Figure 505151: required level of performance and price for each application
Figure 50Figure 51Figure 51 has the purpose to give an idea of which applications might
be more profitable for who provides them. The basic idea is that there are two important
factors: the performance required and the level of price that it is possible to offer. Giving
the evaluation from the questionnaire there are applications that have a high value (more
than 3.5) about performance and for them a provider must to guarantee a high level of
quality (…and so spend much money), other applications don’t require a very high level of
performance (so a provider/operator can provide them without spend a lot of money for
carrying traffic they produce). fromFrom the other side there is the price; there are some
application that don’t have a very severe requirement for low prices (synchronous mirroring
or tele-medicine), so a provider can apply a high mark-up and increase its revenues, for
other applications a low price is a driver for the success.
After this, from an operator point of view the best may be an applicationsapplications that
require low performance (low valuation in the questionnaire) and that doesn’t have severe
requirements about low price (low valuation in the questionnaire).
The histograms presented in Figure 51Figure 52Figure 52 represent the product between
the average performance valuation and the price valuation. As more the histogram is short
as more the related application might be profitable for who provides it.




                                                                                                                                                    Page 167 of 168
                                                                                                                            0
                                                                                                                                2
                                                                                                                                    4
                                                                                                                                        6
                                                                                                                                            8
                                                                                                                                                10
                                                                                                                                                     12
                                                                                                                                                                    14
                                                                                                                                                                              16
                                                                                                                                                                                   18
                                                                                                                 Storage


                                                                                                             - Back-Up /
                                                                                                               Restore


                                                                                                           - Storage on
                                                                                                          demand (SoD)




                                                                                                                                                          Storage
                                                                                                                -
                                                                                                          Asyncrhonous
                                                                                                            mirroring

                                                                                                          - Synchronous
                                                                                                             mirroring




                            point of view
                                                                                                          Grid computing



                                                                                                          - Compute Grid



                                                                                                              - Data Grid




                                                                                                                                                             Grid computing
                                                                                                             - Utility Grid



                                                                                                              Multimedia


                                                                                                            - Video on
                                                                                                                                                                                                     Price X performance




                                                                                                             Demand
                                                                                                          (entertainment
                                                                                                             - Video




Page 168 of 168
                                                                                                          Broadcast (IP-
                                                                                                                                                                                                                                                         European Leadership"




                                                                                                              TV),

                                                                                                               - Video
                                                                                                                                                                                                                                                    Optical network for Broadband
                                                                                                                                                                                                                                                    IST IP NOBEL "Next generation




                                                                                                              Download

                                                                                                           - Video Chat
                                                                                                          (SIF quality, no
                                                                                                             real-time
                                                                                                          - Narrowband
                                                                                                           Voice, data
                                                                                                             (VoIP,...)
                                                                                                                                                                                        Multimedia




                                                                                                          - Telemedicine
                                                                                                           (disgnostic)


                                                                                                                - Gaming


                                                                                                               - Digital
                                                                                                            distribution,
                                                                                                           digital cinema
                                                                                                               - Video
                                                                                                             conference
                                                                                                                (PAL
                  Figure 515252: Evaluation of profitability of each application from operator/provider
                                                                                                                                                                                                                                                                   Title of the document




                                                                                                                                                                                                                                Teil.docD06_DRAFT_V15ejw
                                                                                                                                                                                                                           b2e913b07f59.docD06_DRAFT_V1
                                                                                                                                                                                                                                  e1686d81-2994-497c-93f6-

                                                                                                                                                                                                                                       6JW2 Andreas zweiter

				
DOCUMENT INFO
Shared By:
Categories:
Tags:
Stats:
views:2
posted:9/7/2012
language:English
pages:168