Requirements for Traffic Engineering Over MPLS _draft-ietf-mpls

Document Sample
Requirements for Traffic Engineering Over MPLS _draft-ietf-mpls Powered By Docstoc
					Requirements for Traffic Engineering
           Over MPLS
            (RFC 2702.txt)

          신경망 실험실
          발표자 : 홍 현석

1.0 Introduction
What is MPLS?
Why traffic engineering?
The issues and requirements for
Traffic Engineering on Internet

1.2 Document Organization
Section 2 : the basic functions of Traffic Engineering in the
Section 3 : an overview of the traffic Engineering of MPLS
Section4 : an overview of the fundamental requirements for
Traffic Engineering
Section5 : the desirable attributes and characteristics of
traffic trunks
Section6 : a set of attributes which can be associated with
Section 7 : the introduction of a "constraint-based routing"
framework in MPLS domains
Section 8 : concluding

2.0 Traffic Engineering
mapping traffic flows over network resources

2.0 Traffic Engineering
TE의 목적 및 필요성
   To facilitate efficient and reliable network operations
   Optimizing transmission costs and performance
        Optimization network resource utilization
        To achieve the required QoS including loss and delay
        Spread the network traffic across network links and
         minimize the impact of link failures
        Ensure available spare link capacity for re-routing traffic on
        Also, to meet policy requirements imposed by network
   Rapid changes in demand and marketplace

2.1 Traffic Engineering performance

 Traffic oriented performance objectives
   Enhance the QoS of traffic streams
   In the Best effort service model
    -   minimization of packet loss (most important
    -   minimization of delay
    -   maximization of throughput
    -   Traffic objectives (packet delay variation, loss ratio,
        and maximum packet transfer delay) are useful in
        the differentiated services

2.1 Traffic Engineering performance

    resource oriented performance objectives
     Efficient management of network resources
     Efficient management of bandwidth resources
    Congestion의 원인 (both objectives)
1. When network resources are insufficient or inadequate to
    accommodate offered load.
         해결책
             link 용량 확장
             application of classical congestion control techniques
             both
             Classical techniques : rate limiting, window flow control,
              router queue management, schedule-based control, and

2.1 Traffic Engineering performance

2. Inefficient resource allocation
         causing subsets of network resources to become over-
         utilized while others remain underutilized.
        해결책
                 • Reduced by adopting load balancing policies
        결과
              congestion is minimized
              packet loss decreases
              transit delay decreases
              throughput increases
              Minimize maximum resource utilization
              So, service quality experienced by end users enhanced
   Nevertheless, Flexible capabilities for TE

2.2 Traffic and Resource Control

In the TE process model
   The Traffic Engineer, or a suitable automaton
    acts as the controller
   This system includes
        A set of interconnected network elements
        A network performance monitoring system
        A set of network configuration management tools
   control actions includes (가능한 automaton방법)
        Modification of traffic management parameters
        Modification of parameters associated with routing
        Modification of attributes and constraints associated
         with resources.

2.3 Limitations of Current IGP Control

In router-based cores, TE는 routing metrics의 간
단한 조작으로 행해진다.
                                             The selected path




                   Link metric

2.3 Limitations of Current IGP Control

IGP (interior gateway protocol)
   Shortest path algorithms (SPF)
   Topology driven based on a simple additive metric
   So, bandwidth 나 traffic 특징은 라우팅 결정의 요소가 아님
   AS내에서 어느 정도의 congestion 문제를 해결
Congestion occurs when
   multiple traffic streams converge on specific links or
    router interfaces  sub-optimal resource allocation
   a link or router interface which does not have enough
    bandwidth to accommodate it.
        Solve : Equal cost path load sharing method (not helpful in
         large dense topology, due to the first cause)

2.3 Limitations of Current IGP Control
Mechanisms (overlay model)

TE has been done with layer 2 technologies such
as ATM and Frame Relay
     ATM PVCs

                                 Public switch

                Public switch

                                Public switch

                                ATM Switch

2.3 Limitations of Current IGP Control

A popular approach  Overlay model
   IP over ATM or IP over FR
   Extends the design space by virtual topologies which consist of VC
Overlay model provides
   Constraint-based routing at the VC level
   Higher-speed interfaces and greater aggregate bandwidth
   Explicit VC paths
   Call admission control
   Traffic shaping and policing
Limitations of the IP overlay Model
   More complex network management
        Due to manage two different networks
   N-squared problem due to a full-mesh of ATM PVCs
   Scalability problem

3.0 MPLS and Traffic Engineering
What is Traffic trunk?(5장에서 설명)
Attractiveness of MPLS for TE
(1) Explicit label switched paths ( key point to TE)
(2) LSPs can be efficiently maintained (per-LSP statistics)
(3) Traffic trunks can be instantiated / mapped onto LSPs
(4) A set of attributes associated with traffic trunks modulate their
   behavioral 특성
(5) A set of attributes associated with resources constrain the
   placement of LSPs and traffic trunks
(6) MPLS allows for both traffic aggregation and disaggregation
(7) Aggregation is easy to integrate a "constraint-based routing"
(8) offer significantly lower overhead
(9) MPLS supports for a dynamic protocol, such as RSVP, CR-
   LDP for TE

3.1 Induced MPLS Graph
Is similar to a virtual topology in an overlay
mapped onto the physical network through
the selection of LSPs for traffic trunks
ISSUE : how to efficiently map an induced
MPLS graph onto the physical network

 3.1 Induced MPLS Graph
A capacitated graph
G = (V,E,c)
V : set of nodes
E : set of links
c : a set of capacity
G : “base” network topology
v,w in V and (v,w) is in E and are directly connected under G
Induced MPLS graph
H = (U,F,d)
U : a subset of V
F : the set of LSPs
d : a set of demands and restrictions associated with F
H : Induced MPLS Graph
x,y in U and (x,y) is in F and is an LSP with endpoints
3.2 The Fundamental Problem of
Traffic Engineering Over MPLS

how to map packets onto forwarding
equivalence classes.
how to map forwarding equivalence
classes onto traffic trunks.
how to map traffic trunks onto the
physical network topology through
label switched paths.

4.0 Augmented Capabilities for
Traffic Engineering Over MPLS

A set of attributes associated with traffic trunks
A set of attributes associated with resources
A "constraint-based routing" framework which is
used to select paths for traffic trunks subject to
constraints imposed by items explained above.
Traffic trunks and resources, routing parameters
와 관계된 attributes는 원하는 network state를 만들
기 위해 변경되어짐.

   5.0 Traffic Trunk Attributes and
- Behavioral characteristics
   The basic properties of traffic trunks
       A traffic trunk is an aggregate of traffic flows belonging to the
        same class
            Flows in a traffic trunk are forwarded along a common path
       An abstract representation of traffic to which specific
        characteristics can be associated
            Flows in a traffic trunk share a common QoS requirements
       At the current internet, a traffic trunk could encapsulate all of
        the traffic
       Traffic trunks are routable objects (similar to VCs)
       A traffic trunk is distinct from the LSP
       Unidirectional
   Two basic issues
       Parameterization of traffic trunks
       Path placement and maintenance rules for traffic trunks

5.1 Bidirectional Traffic Trunks
One trunk : forward trunk
The other trunk : backward trunk
Two conditions
    Both traffic trunk are instantiated through an atomic
     action by one LSR or a network management station
    Both are instantiated and destroyed together
Topological properties
    Topologically symmetric : component traffic trunks are
     routed through the same physical path
    Topologically asymmetric : traffic trunk routed through
     the different physical paths

5.2 Basic Operations on Traffic Trunks

Establish : to create a traffic trunk
Activate : to cause a traffic trunk to start passing
traffic(implemented one atomic action)
Deactivate : to stop traffic
Modify Attributes
Reroute : a traffic trunk to change its route
Destroy : to remove a traffic trunk from the

5.3 Accounting and Performance

Are very important to the billing and traffic
characterization functions
Performance statistics can be used for
traffic characteristics, performance
optimization and capacity planning

5.4 Basic Attributes of
Traffic Trunks
Traffic parameter attributes
Generic Path selection and
maintenance attributes
Priority attribute
Preemption attribute
Resilience attribute
Policing attribute

5.5 Traffic Parameter Attributes

Characteristics of the traffic streams
Peak rates, average rates,
permissible burst size, useful for
resource allocation and congestion

 Traffic Trunk example

Flows   Trunks


Per-PVC statistics
 Monitoring of traffic patterns for optimal
  PVC placement and management
 Provides the information needed to
  modify the virtual topology under