Get documents about "atm
Shared by: changcheng2
-
Stats
- views:
- 62
- posted:
- 11/22/2011
- language:
- English
- pages:
- 31
Document Sample
VARIOUS APPROACHES FOR IP OVER ATM
Abid Akbar
Department of Computer Science
California State University, Chico
400 W. First Street,
Chico, CA 95929
abid@ecst.csuchico.edu
ABSTRACT
ATM is emerging as primary networking technology for next generation
multimedia communications. ATM delivers important advantages over existing
LAN and WAN technologies including the promise of scalable bandwidth and
quality of service (QoS) guarantees. Due the popularity and dominance of IP, the
success of ATM as a data networking technology lies largely in its ability to
support IP traffic on top of it. This paper is a study of various approaches adopted
and current work in progress for transporting IP data traffic over ATM. It
examines various pros and cons of these approaches.
1. INTRODUCTION
The construction and deployment of Asynchronous Transfer Mode (ATM) network is a recent
development in the field of computer communications. The success of ATM lies largely in its
ability to transport legacy data traffic, mostly IP, over its network infrastructure. Integrating this
new technology into the existing Internet requires schemes for managing the transmission of IP
datagrams over ATM networks. Such schemes will take advantage of the strengths of ATM
while effectively bridging the gap between the data forwarding models of ATM and the Internet.
1
2. BACKGROUND
2.1 ASYNCHRONOUS TRANSFER MODE (ATM)
ATM is emerging as primary networking technology for next generation multimedia
communications. It is clear that ATM will play a central role in the evolution of current
workgroup, campus and enterprise networks. ATM combines the reliability of circuit switching
with the efficiency of packet switching, providing best way to deliver all types of data.
ATM is a connection-oriented service that transfers small, fixed size packets called cells through
a switched based network. Network protocols called ATM Adaptation Layers (AALs) fragment
larger, variable sized packets into cells for transmission and reassemble them upon arrival at
their destination.
ATM technology is designed to meet the needs of heterogeneous high-speed networking. ATM
is currently gaining popularity, however it is uncertain whether or not ATM will become a
dominant networking technology. The existing installed base of LANs such as Ethernets is
considerable; replacing these networks with ATM will be costly and in some cases unnecessary.
In near future, it appears that ATM networks will be used as backbones connecting existing
LANs.
In the current Internet, the solution to forwarding data through such a heterogeneous
internetwork is provided by the Internet Protocol (IP). IP is almost entirely independent of the
subnet technology used. [1] & [3]
2
2.2 INTERNET PROTOCOL (IP)
The Internet Protocol performs two primary functions.
- Determining a route and relaying packets across the Internet.
- Segmentation of packet, if necessary, to accommodate a network that has a small
maximum packet size and then reassembly of packets when they reache the destination. IP
is almost entirely independent of the subnet technology used, it makes few assumptions
about the nature of individual subnets. IP packets can traverse many different types of
subnets (including ATM networks) without either the sender or receiver being aware of the
details of the network encountered along the path. Unlike ATM, IP is a datagram and does
not require the establishment of connection before data can be sent.
3. THE CHALLENGES OF IP OVER ATM
3.1 CONNECTION ORIENTED VS CONNECTIONLESS
ATM is a connection-oriented service that means a connection needs to be established between
two parties before they can send data to each other. Once the connection is setup. All data
between then is sent along the connection path. On the contrary, IP is connectionless service, that
means no connection is needed and each IP packet is forwarded by routers independently on hop
by hop basis. The attitude of IP can be characterized as “send and forget”. To transport IP traffic
over ATM two options can be considered.
-A new connection is established on demand between two parties
-Data is forwarded through pre-configured connection
In 1st option, if the amount of data to be transferred is small, then the cost of setting up and
3
breaking the connection can not be justified. In 2nd option, this can happen that the pre-
configured path may not be an optimal path and it may become overwhelmed by the amount of
data being transferred.[2]& [3]
3.2 QOS Vs BEST EFFORT
Quality of service is an important concept in ATM networks. ATM networks have the potential
to provide real time performance guarantees such as bounds on bandwidth and packet loss. These
performance guarantees are necessary for many network applications, such as digital audio and
video. IP, on the other hand, has no such concepts and each packet is forwarded on best effort
basis by routers. To get benefit of the QoS guarantees of ATM networks, IP needs to be modified
to include that information.
As ATM and Internet will likely co-exist in the near future, it is desirable that hosts using these
two types of networks be able to exchange data. One approach is to use ATM network as a
datalink layer, similar to Ethernet. This approach is commonly referred to as IP over ATM. [2] &
[3]
TCP UDP Data Link Layer
IP Network Layer
AAL Transport Layer
Ethernet FDDI ATM
4
4. MODELS OF IP OVER ATM
To run IP on top of ATM network, it is necessary to figure out how to relate ATM protocol
layers to TCP/IP protocol layers. Various models have been considered for this purpose and
presented here.
In classical model, an ATM network or internetwork is used as a subnet in an IP internetwork.
ATM attached hosts establish virtual circuits between themselves to carry IP datagrams. An
Address Resolution Protocol (ARP) server handles the translation between IP addresses and
ATM addresses. ATM network can be seen as an opaque routing cloud. Routers at the edge of
ATM network (or somewhere else on the Internet) are oblivious of its internal details and
topology. Traffic between hosts in different logical subnets has to go through a router even
though they are attached to the same ATM network. This is not desirable since routers introduce
a high latency and become a bandwidth bottleneck. Next Hop Resolution Protocol (NHRP) is
used to solve this problem.
A 2nd approach uses an ATM network to simulate LAN protocols like Ethernet or Token Ring.
IP runs on top of ATM network in same way it runs on top of Ethernet or Token Ring. This is
known as LAN Emulation. LANE allows current IP applications to run over ATM network
without modification. LANE can accelerate the deployment of ATM network. However, traffic
between different ELANs (Emulated LANs) still needs to travel through router. A combination
of LANE and NHRP called Multi Protocol Over ATM (MPOA) solves the problem by creating
routes that bypass routers between ELANs.
5
A 3rd approach proposes connectionless servers within ATM networks to handle datagrams such
as those generated by IP. In this arrangement hosts establish virtual circuits to connectionless
servers within ATM network. Connectionless servers forward datagrams in much the same way
as IP routers. Virtual circuits between connectionless servers carry datagrams towards their
destination. This approach has the benefit of multiplexing virtual circuits, but due to this sharing,
it has difficulties providing performance guarantees or protection between competing traffic
sources. [4] & [5] For this reason this model is not included in this paper for discussion. Model is
shown in Figure below
Host
Host
CLS Internet
Internet
ATM Network R
R
CLS
Host
Host
CLS = Connectionless Server, R = Router
5. CLASSICAL IP OVER ATM (CIOA)
A configuration of CIOA over ATM is shown in figure.
Router 1 Router 2 Router n - 1 Router n
R R R R
LIS1 LIS2 LISn-1 LISn
LIS = Logical IP Subnetwork
6
In classical IP over ATM model, the ATM fabric interconnecting a group of hosts is considered a
network, called Non broadcast Multi Access (NBMA). An NBMA network is made up of a
switched service such as ATM or frame relay, with a large number of hosts that cannot directly
broadcast messages to each other. An NBMA network is subdivided into several logical IP
subnetworks (LIS).
Hosts in LIS share the same IP prefix and address mask. So LIS is much similar to a traditional
IP subnetwork. The major difference is that traditional IP subnets are separated from each other
by routers, while LISs are actually connected to same ATM network. This is the reason why
these subnets are called logical subnets. Software configuration defines the logical subnets and it
has nothing to do with hardware setup. Moreover it is obvious that inter LIS communication
need not necessarily go through a router.
Hosts in a LIS communicates with each other through end to end ATM connections.
If a host A needs to communicate with host B, which is in the same LIS, 1st it has to establish a
connection with B. A has B's IP address but does not know its ATM address. In order to resolve
IP addresses to ATM addresses, each LIS contains an Address Resolution Protocol (ARP) server
which is called ATMARP server. Server provides ATM address to A in response to its query. A
can then establish a connection through P-NNI signal. Hosts in different LISs communicate with
each other through routers. [4], [9] & [10]
7
5.1 DRAWBACKS OF CLASSICAL IP OVER ATM
Reliability of ATMARP server is questionable. If the server suffers a catastrophic failure, all
hosts on LIS would be unable to use ARP. Moreover in CIOA over ATM each host needs to be
manually configured with the ATM address of ARP server.
Router forward IP traffic between subnets. Each LIS contains a router and IP packets that are for
a host in another LISs are forwarded to the router. Router forwards this packet to another router
and packet is routed to the destination on hop by hop basis. This is not desired in actual practice
since each router has to reassemble and disassemble the IP packet and this introduces a huge
amount of delay. It is more suitable to establish an end to end direct connection between two
communicating parties as they are attached to the same ATM network, hop by hop forwarding is
definitely a waste of time and resources. Next Hop Resolution Protocol (NHRP) fixes this
problem by allowing direct connection between the hosts that lie in different LISs.
5.2 ATM ADDRESS RESOLUTION PROTOCOL (ATMARP)
In order to operate IP over ATM, a mechanism must be used to resolve IP addresses to their
corresponding ATM addresses. In order for router to pass a packet across an ATM network, it
must have an address resolution table to determine the ATM address of the destination next hop
router. Address resolution table can be configured manually. However, a better approach is to
use CIOA protocol which supports an automatic address resolution of IP addresses. This protocol
applies the concept of LIS
8
To resolve the addresses of hosts within the LIS, each LIS supports an ATM ARP server. All
hosts within the LIS are configured with the unique ATM address of the ATM ARP server.
When a host comes up within the LIS, it first establishes a connection to the ATM ARP server
using the configured address. Once the ATM ARP server detects a connection from a new LIS
client, it transmits an inverse ARP REQUEST to the attaching client and requests the host IP and
ATM addresses. These addresses are stored in ATM ARP table by the server. Subsequently, any
host within the LIS, which wishes to resolve a destination IP address, sends an ATM ARP
request to the server, which then responds with an ATM ARP reply if an address mapping is
found; if not, it returns an ATM_NAK response to indicate lack of registered address
mapping.[7] 144.254.10.2
R
2 ARP Server
Address Resolution
1 144.254.10.2 A
.3 B
Routing
144.254.10.x Direct
.23.x via 144.254.10.2
.45.x .3
.67.x .3
ATM Network
3
Host R R
144.254.10.3
144.254.45.9 4
Step 1: Routing table maps final destination
Step 2: Address resolution table or server maps next hop IP address to ATM address
Step 3: Signaling creates ATM virtual connection between routers
Step 4: Forward packet over ATM virtual connection
5.3 DATA ENCAPSULATION
Data Encapsulation is an important aspect to transfer any network layer protocol (IP) over an
overlay mode (ATM network). Communication between two devices requires either that two
9
devices agree on a common form of encapsulation or that an internetworking device (e.g. router)
be used to convert between two forms of encapsulation. Two modes of encapsulation can be
considered, VC Based Multiplexing and LLC/SNAP Encapsulation. [7]
5.3.1 VC BASED MULTIPLEXING
In it a single protocol is carried across an ATM connection. The type of protocol is implicitly
identified at connection setup. As a result, no multiplexing or packet type field is required or
carried within the packet. This approach is preferred when large number of VCs can be
established in a fast and economical way.
5.3.2 LLC/SNAP ENCAPSULATION
With LLC encapsulation, number of protocols can be carried over same VC connection. In this
approach, an IP packet will be prefixed with an IEEE 802.2 LLC header before it is encapsulated
into the AAL5 frame. This approach is suitable when a separate VC for each carried protocol is
either expensive or not possible.
The LLC/SNAP encapsulation is the most common encapsulation used in the IP over ATM
protocols at this time. The ITU-T has also recently adopted this as the default encapsulation for
multiprotocol transport over ATM. It is used as default encapsulation method for all IP over
ATM protocols.
10
5.4 NEXT HOP RESOLUTION PROTOCOL (NHRP)
In case of CIOA, inter LIS communication has to go through routers. This is not an optimal
solution especially when both parties involved are attached to the same ATM network. It is
desired to establish a direct connection between them and this is not difficult to achieve. What
required, is a mechanism for an host to resolve the IP address of another host (in a foreign LIS)
into its corresponding ATM address. NHRP performs this task.
NHRP consists two types of entities. NHRP server (NHSs) and NHRP client (NHC) and
protocols between them. Each host is NHRP client and each LIS contains at least one NHRP
server. When a host needs to resolve an IP address, it sends a request to NHRP server which is
incharge of its LIS . A NHS can serve more than one LISs and its keep a table of <IP address,
ATM address> pairs for all the hosts that belong to the LISs it is serving. If a pair that matches
with IP address is found, the corresponding ATM will be returned, otherwise -ve reply will be
returned.
So NHS behaves similar like ATMAPR server. In practice LIS where NHS and ATMAPR
clients co-exist, NHS is coupled with the function of ATMAPR server. The only limitation of
ATMAPR server is that it can not resolve an IP address that belongs to another LIS while NHRP
server can do that. When a query comes to NHRP server regarding the IP address that belongs to
a LIS it does not serve, it will manage to forward the query to the NHRP server that serves that
LIS.
NHSs that serve LISs on ATM network have pre-configured connections between them, so that
11
they form a routed network for queries. Routing protocol OSPF executed among these NHSs
helps NHSs know which next hop (another NHS) to forward the query in order to reach the
destination NHS. When the NHS that serves the LIS receives the query, it will reply with
corresponding ATM address to the end system that initiates the query. This reply will travel back
through the intermediate NHSs, these NHS may cache the <IP address, ATM address> entry, so
that next time NHRP queries with same IP address can be intercepted and replied. This feature
saves lot of NHR traffic. Once a sender knows the ATM address of the receiver, it is possible to
establish an end to end connection with receiver (shortcut) to transfer IP packets between them.
Router 1 Router 2 Router 3 Router 4
R R R R
A LIS1 LIS2 LIS3 LIS4 LIS5 B
NHS1 NHS2 NHS3 NHS4 NHS5
LIS = Logical IP Subnetwork, NHS = Next Hop Resolution Protocol Server
Figure shows an ATM network that is partitioned into five LISs. Each is served by separate
NHS. Router1 is connected to LIS1 & LIS2, router2 is connected with LIS2 & LIS3, router3 is
connected with LIS3 & LIS4 and router 4 is connected with LIS4 & LIS5. Hosts A & B are
attached to LIS1 and LIS5 respectively. If A wants to send data to B, in classical IP over ATM,
data will have to travel through router 1, 2, 3 and 4. With NHRP, A will send NHRP query to
12
NHS1, it will be forwarded to NHS2 then to NHS3, NHS4 and ultimately to NHS5. Since NHS5
is serving LIS5, which contains the host B, NHS5 will reply to A with ATM address of B. When
the reply travels back, it may go through NHS4, NHS3, NHS2 & NHS1, each of these will cache
the information so that future request will be replied directly by them without forwarding it to the
deriving NHS. When A gets the reply, it establishes an end to end connection with B to transfer
the data between them. [9]
5.5 IP MULTICAST OVER ATM
Classical IP over ATM and NHRP only support IP unicast over ATM. To support IP multicast,
two issues need to be resolved. First a need for an address resolution protocol to translate
multicast IP address into a list of ATM addresses, and this is solved by Multicast Address
Resolution Server (MARS). Second, a need to specify how multicast data is transferred among
the involved parties. VC Mesh and Multicast Server (MCS) are two possible solutions.
A Multicast Address Resolution Server is introduced into each LIS to perform the multicast
address resolution. It answers the queries for multicast addresses from the hosts in the same way
as ATMAPR server answers queries for unicast addresses. A host leaves or joins a particular
multicast group by sending Internet Group Multicast Protocol (IGMP) packets to the MARS.
When a multicast IP address is resolved into a list of ATM addresses, the data needs to be
forwarded among the group members from the sender to the receiver. One way to do this is to let
each group member set up a point-to-multipoint connection with all other group members and
this approach is called VC Mesh. The other way is to introduce a Multicast Server (MCS) into
13
each LIS that supports multicast. When a host makes query for a multicast address, MARS will
reply to it with the ATM address of the MCS. The host then sends the multicast packets to MCS.
The MCS will build a point-to-multiple point connection or multiple point-to-point connection to
the group members to forward the packet received from the host to all the members of the group
specified in the address field of the multicast packet.
VC mesh and MCS each has its pros and cons. With MCS, if the membership of a multicast
group changes, it only needs to modify the point-to-multipoint VC to the group members while
with VC mesh, all connections in this "mesh" have to be modified. However, MCS needs to
reassemble the cellified packets sent from the source and resend them to the group members so it
may become the single point of congestion and introduce certain amount of latency. With VC
mesh the reassembling is not needed so the latency is minimized. [6]
6. LAN EMULATION (LANE)
In LAN Emulation (LANE), an ATM network is configured to simulate an Ethernet or Token
Ring LAN – although operating at higher speed than a real such network. The resulting LAN is
called an Emulated LAN (ELAN).
The motivation for this approach is that it requires no modifications to higher layer protocols to
enable their operation over an ATM network. By emulating the behavior of legacy networks,
LANE provides support to ATM users faced with the problem of interconnecting their installed
base of LAN protocols over a new ATM medium, while at the same time maximizing the impact
to their existing systems.
14
LANE features can be summarized as follows
1. LANE provides a mechanism for existing LAN based client / server applications to run
over ATM networks without modifications
2. LANE uses ATM as a baseband to interconnect existing legacy LANs to achieve higher
bandwidth.
3. LANE permits several emulated LANs to concurrently share the same ATM network.
This allows one physical network to appear as several logical networks.
One of the chief benefit of LANE is the ability of all devices attached to a LANE network to
function in a plug and play fashion, requiring minimum configuration.
LANE currently is very functional, but LANE protocol suite is evolving, it will continue to be
developed and enhanced for years to come.
LANE protocol defines a service interface for higher layer (that is for network layer) protocols,
that is identical to that of existing LANs. It enables data sent across the ATM network to be
encapsulated in the appropriate LAN MAC packet format. In this way the IP software that is
running previously on Ethernet and token ring can be ported onto the ATM network without any
modification. This helps accelerate the deployment of ATM as a LAN technology. The LANE
protocol supports a range of Maximum Protocol Data Unit (MPDU) sizes. However all
emulation clients (LEC) within a given ELAN must use the same MPDU size. Figure shows
traditional LAN Vs emulated LAN [7], [6] & [9]
15
LAN Hub Physical Medium
LAN Hub
REAL LAN
Routers
User Servers R
ATM Network EMULATED LAN
Routers
Users Servers R
The Basic function of the LANE protocol is to map MAC addresses into ATM addresses. Figure
shows a protocol model.
LES1
LNN
I
Example of target LESn
LEC ELAN
UNI
ATM Host
ATM Network
Desired Connectivity (LANE BUS1
Service)
Layer
2 LNN
Switch I
Ethernet
BUSn
LECS
R
Ethernet
LUN
LES: LAN Emulation Server
I
LECS: LAN Emulation Configuration Server
LEC: LAN Emulation Client
BUS: Broadcast and Unknown Server NMS
LUNI: LAN Emulation User to Network
Interface
LNNI: LAN Emulation Network Node Interface
UNI: User Network Interface
NMS: Network Management System
LANE Protocol Model
16
6.1 LANE ELEMENTS
LANE specifies following four types of entities and protocols. [8]
6.1.1 LAN EMULATION CLIENT (LEC)
A LEC runs on each ATM host in an ELAN to simulate an Ethernet or Token Ring node. Each
LEC has one or more MAC addresses associated with it. It contacts the LES to resolve the MAC
addresses into ATM addresses and performs certain control functions. It emulates Ethernet or
Token Ring service interface to the IP layer by encapsulating the outgoing IP packets into ELAN
frames or decapsulating the incoming ELAN frames into IP packets.
6.1.2 LAN EMULATION SERVER (LES)
Each ELAN contains a LES, which acts as the coordinator. Each LEC will register with LES its
<MAC address, ATM address> pair. Based on such information LES resolves MAC addresses
into corresponding ATM addresses in the same as ATMARP server does in Classical IP over
ATM.
6.1.3 BROADCAST AND UNKNOWN SERVER (BUS)
Each ELAN includes a Broadcast and Unknown Server (BUS) to emulate the broadcast
capability of Ethernet and Token Ring. A LEC who wants to broadcast a packet sends it to BUS,
which forwards every ELAN member a copy. Before the direct data connection is built between
17
two LECs, the data between them is also forwarded through BUS.
6.1.4 LAN EMULATION CONFIGURATION SERVER (LECS)
There can be more than one ELAN running on an ATM network. LECS keeps the configuration
information of each ELAN including the LECs, LES and BUS in each ELAN.
6.2 OPERATION OF LANE
6.2.1 CONFIGURATION
In this step, a LEC contacts the LECS to know each ELAN and the address of LES and BUS to
contact to join a particular ELAN. There are three ways for a LEC to access LECS. The first is to
configure the ATM address of the LECS into the LEC. The second is to have a fixed VPI/VCI
that directs to the LECS from every end system. The third is to get it through ILMI.
6.2.2 REGISTRATION
After knowing the ATM address of the LES of a particular ELAN, LEC sends a registration
message to the LES, which includes the MAC address and ATM address of the LEC. LES after
receiving the message records the information in its address resolution table and creates
connections to the LEC for the transfer of data and control information.
6.2.3 BUS CONNECTION
18
LEC establishes a connection with BUS by using its ATM address obtained at the configuration
stage, a connection to the bus is established for the transfer of multicast data.
6.2.4 DATA TRANSFER
When LEC wants to send some data to another LEC, it obtains ATM address of another LEC
from LES. After obtaining the address it establishes a direct data connection with that LEC.
Ethernet or Token Ring frames between them are transported on this connection in AAL 5
frames with LLC encapsulation. If LEC needs to broadcast a packet, it simply sends the packet to
the BUS, which forwards every member of ELAN a copy of the packet.
6.3 DRAWBACKS OF LANE
The disadvantage of LANE is that it hides ATM features from higher protocols. It means that
any network layer protocol that operates over ATM through LANE cannot benefit from the QoS
properties of ATM. Although most applications today do not expect to receive and do not request
any guaranteed QoS from the underlying network protocol, however this situation will likely
change in future. Considerable work is being done on building a networking infrastructure
capable of supporting a new class of multimedia applications that combine voice, video, image
and data traffic. To support such traffic, QoS guarantees will be required from the network.
LANE also suffers from the drawback that by definition it behaves like protocol independent
bridging. Bridging is effective for interconnecting small workgroups, but it does not scale well to
support large networks.
19
LANE only supports emulation of one type of network at a time. If a host on an emulated
Ethernet wants to communicate with a host on an emulated Token Ring, the packets must pass
through a router that is a member of both emulated LANs.
7. MULTIPROTOCOL OVER ATM (MPOA)
LANE Emulation and Classical IP over ATM are only starting point for building ATM networks.
In order to take full advantage of ATM’s potential, new paradigms in network design and
application development must be undertaken. MPOA is a product of this paradigm shift and may
revolutionize the way the networks are built and used.
The first order of business for MPOA is to ensure that both bridging and routing are preserved
for legacy LANs and the VLAN topology in use.
An MPOA network uses LAN for bridging function. An ELAN scope is a single layer 3 subnet,
where as MPOA is concerned with subnet connectivity. For layer 3 forwarding function MPOA
adopts and extends NHRP. So MPOA is a combination of LANE and NHRP. MPOA improves
LANE by allowing inter ELAN traffic to go through shortcut connections rather than through
routers. In order to build such a connection, NHRP is used to resolve destination IP address into
ATM address. So MPOA is a combination of layer 3 routing and layer 2 bridging.
If one legacy system attached to an edge device needs to send data to another legacy system
attached to another edge device, the best approach is to establish a direct connection between the
20
two edge devices and transport traffic across this connection. The edge device that the sender is
attached to is called an ingress endpoint and the edge device the receiver is attached to is called
an egress end point. MPOA is to build end-to-end connection between an ingress endpoint and
an egress endpoint for efficient communication as well as enabling applications to make use of
network’s ability to provide guaranteed QoS. [9] & [6]
7.1 MPOA ADVANTAGES
- Clients can establish direct connections to remote servers without always having to transit
through routers
- Lower latency in establishing connections between devices
- Reduced amount of broadcast traffic
- Flexibility in selection of maximum transfer unit size to optimize performance
7.2 MPOA VS LANE
- MPOA is an evolution of LANE; it uses LANE
- LANE operates at layer 2: bridging
- MPOA operates at layer 2 and layer 3: bridging and routing
- LANE hides ATM/QoS, MPOA exposes both
- LANE requires no modifications to host protocol stacks, MPOA requires modification
7.3 VIRTUAL LAN & MPOA
21
One of the MPOA’s main goals is to enhance the ability of network designers to build a single
physical ATM network for enterprise networks, while at the same time, providing the ability to
subdivide the network along the administrative boundaries (i.e. build virtual LANs). The concept
of virtual LAN is a generic idea, where the hosts logical addresses are detached from their
physical locations.
Virtual LANs can divide the network into a group of hosts and can restrict the access of these
groups have with the servers. In this way virtual LAN acts as a firewall to provide additional
security.
Virtual LANs are limited by the security of the managing database or the facilities the network
layer protocols provide by their support of MPOA servers. The major drawback to the current
techniques used to build virtual LANs is that they are not standard-based and are potentially
limited in how large the total system can become. MPOA, on the other hand, provides the
functionality with standard-based technology and is not limited with regard to scalability.
7.4 APPROACHES
Three different approaches are presented for MPOA.
7.4.1 PEER APPROACH
22
Various schemes have proposed peer approach. These schemes, however, include various
proprietary IP switching schemes. Such schemes generally do not support VLANs and hence
apply in different parts of the network than MPOA.
7.4.2 INTEGRATED P-NNI APPROACH
This model proposes that the P-NNI protocol to be used by both ATM switches and packet
routers. This is because P-NNI protocol is more powerful and scalable routing protocol than any
other protocol existing in current routed networks. There are significant open issues, however,
about how viable it may be to map a connection oriented routing protocol like P-NNI -- with its
high processing requirements and large latencies – to connectionless devices like router.
Obtaining reasonable packet forwarding rates on routers running P-NNI will likely preclude the
use of the QoS capabilities of P-NNI, hence eliminating one of the great attractions of P-NNI.
The ATM forum currently has a working group on P-NNI that is considering such issues.
7.4.3 DISTRIBUTED ROUTER PROTOCOL APPROACH
Recent work on MPOA is based on new vision of VLANs, which extend beyond the first
generation of LANE based VLANs. First generation of VLANs are built around layer 2 LAN
switches and support the LANE protocol. The first generation, however, suffers from two
problems: the bottleneck of requiring router hops for virtual LAN interconnection and the
inability to run protocols in native mode, thereby exploiting the QoS features of ATM.
Beyond this first generation, number of companies plans to develop a new generation of layer 3
LAN switching systems including Cisco systems. Such switches would act not as simple bridges
23
but would also switch packets based on their network layer addresses and other higher layer
attributes. In essence, a system of such layer switches would constitute a distributed router.
Layer 3 based VLANs would provide number of advantages over LANE based layer 2 VLANs.
7.5 MPOA OPERATION COMPONENTS AND ARCHITECTURE
An MPOA network consists of several network layer-aware components which can be
subdivided into router servers, edge connection devices, LANE servers, NHRP servers and ATM
attached hosts. In the MPOA model, the ATM fabric is considered to be one physical network
capable of supporting many virtual LANs. Each network while separate is reachable using
shortcuts. In a sense, ATM network can be visualized as an emulated multiprotocol bridge router
with the addition of very high bandwidth capabilities.
The key architectural components are as follows:
- Edge Connection Devices are used to physically attach legacy networks to an MPOA
system (i.e. an Ethernet to an ATM converter). These are similar to LANE bridges.
- Route Servers have topological information gained by running routing protocols and
distributing state among themselves. These are network components that support MPS
functions along with LANE and NHRP.
- Information flows comprise the protocol descriptions for MPOA Client (MPC) to MPOA
Server (MPS), MPC to MPC, and MPS to MPS exchanges.
7.6 INFORMATION FLOWS
24
An MPOA system utilizes several information flows. The information flows describe how the
components exchange MPOA state information and resolve target addresses and state. The
MPOA system works by allocating tasks to groups and then defining the protocol’s operation by
specifying information flows among the groups.
To define information flows between functional groups, problem can be broken down into
specific cases corresponding to different states of the protocol. MPOA’s logical components can
be divided into clients and servers, so the implementation of protocol follows four steps.
- Configuration
- Discovery
- Address resolution
- Data transfer
7.6.1 CONFIGURATION
Before MPC and MPS can begin using the MPOA system, they must be registered and
configured. The configuration process is accomplished either manually by a network
administrator or the MPC/MPS can make use of a LANE configuration server. As in LANE,
devices in an MPOA network usually contact the configuration server at boot time. The
configuration server knows which clients and servers are associated within which virtual
networks and the configuration server notifies the clients and servers of their respective
MCS/MPS ATM address.
25
When route servers are initialized, they pass a TLV identifying themselves to the LECS
specifying a configuration request. They are given the identity of the subnetwork(s) they control
along with layer 3 protocol type(s) used. In addition, the route server is a member of the
subnetwork(s) so it also acquires layer 3 address.
The MPCs then register with the LECS by sending a configuration request containing a TLV
identifying the MPC. When MPCs and ATM attached hosts are initialized on an MPOA network
via the LECS, they are given information about which policies should be followed for shortcut
setup, when to time out and delete idle virtual circuits, and which protocol should be using
shortcuts.
7.6.2 DISCOVERY
It is concerned with the set of information exchanges used to inform each MPOA device of its
existence, capabilities and domains. The term discovery when related to MPC, describes the
ability to determine the location of the NHS. As with configuration phase, the mapping from
network layer address to ATM address is done via LANE LE_ARP. Once the discovery phase is
complete, the MPOA components within a domain can pass NHRP/LANE messages. MPC can
now communicate among themselves across subnetwork boundaries.
7.6.3 ADDRESS RESOLUTION
26
Once a host has been configured and has registered itself on the network, it can begin to
communicate with other hosts. In order to communicate, the mapping of layer 2 to layer 3
address must be resolved via the MPS. Data flow between computers on an MPOA system can
be one of three types.
-Intrasubnetwork via the LANE servers
-Intrasubnetwork via default forwarding
-Intrasubnetwork via shortcut routing
7.6.4 DATA TRANSFER
When MPC has successfully received a response to its NHRP query, it can establish a direct
virtual circuit and begin transferring user data. Several base rules have been specified to ensure
smooth default operation. MPOA specifies that as a baseline the parameters documented in RFC
1755 covering signaling parameters for Classical IP over ATM should be used for user data
communication.
Virtual circuits used for data communication or to pass control messages can be deleted when
their usefulness is no longer apparent.
8. CONCLUSION
27
Recent trends in networking (particularly high bandwidth and switched-based networks) have led
to several new networking technologies that are candidate for both LAN and WAN networks.
ATM is one such technology that is gaining acceptance in both industry and research fields. One
issue of particular importance is how to use ATM networks as a portion of the Internet. The
differences between ATM networks (virtual circuits, with possibility of performance guarantees)
and Internet (datagrams, with best effort service) create some interesting, new challenges for
research. As ATM and the Internet will likely co-exist in the near future, it is desirable that hosts
using these two types of networks be able to exchange data.
The paper has presented various approaches currently being adopted for transporting IP data over
ATM. Classical IP Over ATM is easy to implement. However the drawback of it is that inter LIS
traffic has to travel through a router even though both parties are directly connected to the ATM
network. NHRP fixes this problem by augmenting it with an address resolution protocol so that
shortcut connections can be established between end systems that belong to different LISs. To
accelerate the deployment of ATM technology, LANE emulates Ethernet and Token Ring LANs
on an ATM network so that existing IP software running on such LANs can run on ELANs
without modification. However, ELAN suffers the same drawback as Classical IP Over ATM,
that is, inter ELAN traffic has to travel through a router. MPOA combines LANE and NHRP
technology to support both IP routing and LAN bridging over an ATM network.
All these techniques have their pros and cons. Moreover current work is in progress to find a
better approach.
28
9. GLOSSARY
AAL ATM Adaptation Layer
ARP Address Resolution Protocol
ATM ARP ATM Address Resolution Protocol
ATM Asynchronous Transfer Mode
BUS Broadcast and Unknown Server
CIOA Classical IP Over ATM
ELAN Emulated LAN
FDDI Fibre Distributed Data Interface
IGMP Internet Group Multicast Protocol
IP Internet Protocol
ITU-T International Telecommunication Union – Telecommunication
LAN Local Area Network
LANE LAN Emulation
LEC LAN Emulation Client
LECS LAN Emulation Client Server
LES LAN Emulation Server
LIS Logical IP Subnetwork
LLC Logical Link Control
MAC Medium Access Control
MARS Multicast Address Resolution Server
MCS Multi Cast Server
MPC Multimedia PC
29
MPDU Maximum Protocol Data Unit
MPOA Multi-Protocol Over ATM
MPS MPOA Server
NBMA Non Broadcast Multi Access
NHC NHRP Client
NHRP Next Hop Resolution Protocol
NHS NHRP Server
OSPF Open Shortest Path First
P-NNI Private Network Node Interface
QoS Quality of Service
SNAP Subnetwork Access Protocol
TCP Transmission Control Protocol
TLV Type Length Value
UDP User Datagram Protocol
VC Virtual Circuit
VCI Virtual Channel Identifier
VLAN Virtual LAN
VPI Virtual Path Identifier
10. REFERENCES
[1] http://www.npac.syr.edu/users/dpk/ATM_Knowledgebase/ATM-technology.html
[2] http://www.ietf.org
[3] http://www.atmforum.com
30
[4] Bruce, A., “On the Use of QoS in IP Over ATM”, The Tenet Group, Computer Science
Department, University of California at Berkley, Internet Posting
[5] Firoiu, V., Kurose, J., and Towsley, D., “Performance Evaluation of ATM Shortcut
Connections in Overlaid IP/ATM”, Department of Computer Science, University of
Massachusetts, Internet Posting
[6] Minoli, D. and A. Schmidt, “MPOA over ATM, Building State of the Art ATM Internets”,
Manning Publication Co., New Jersey, 1998
[7] Alles, A., and Minoli, D., “LAN, ATM and LAN Emulation technologies”, Artech House
Inc., MA, 1996
[8] http://www.iphase.com/docs/whitepapers/lanemul.cfm
[9] Alles, A., “ATM Internetworking”, Internet Posting, Cisco Systems, 1995
[10] M. Laubach, “Classical IP and ARP Over ATM”, Internet Request for Comment 1577, Jan
1994
31
