This chapter describes how to configure a NETBuilder II bridge/router to establish LAN, WAN, and MAN connectivity through Asynchronous Transfer Mode (ATM) with LAN emulation.
For conceptual information, see "How ATM and LAN Emulation Work" on page 183.
This section describes how to configure your bridge/router to transmit and receive data over an ATM interface using LAN emulation. Two procedures are provided: one for setting up an Ethernet configuration and one for setting up a Token Ring configuration with transparent bridging. See "Setting Up LAN Emulation Client Source Routing" on page 179 for a procedure for setting up a Token Ring network for LAN emulation client source routing.
For detailed descriptions of all commands, see Reference for NETBuilder Family Software.
Before beginning this procedure, complete the following tasks:
To perform LAN emulation over an ATM network, see Figure 56 and follow these steps:
Figure 56 Enabling a Port for LAN Emulation on Ethernet
1 . Specify the emulated LAN name using:
SETDefault !<vport> -ATMLE ElanName = "<string>"
SETDefault !v1 -ATMLE ElanName = "elan1"
2 . Create a virtual port for a LAN emulation client (LEC) to be attached to the ATM network using:
ADD !<port> -PORT VirtualPort <path> ETHATM
ADD !v1 -PORT VirtualPort 4 ETHATM
To perform Token Ring LAN emulation over an ATM network, see Figure 57 and follow these steps:
Figure 57 Enabling a Port for LAN Emulation on Token Ring
1 . Specify the emulated LAN name using:
SETDefault !<vport> -ATMLE ElanName = "<string>"
SETDefault !v2 -ATMLE ElanName = "elan2"
2 . Create a virtual port for a LAN emulation client (LEC) to be attached to the ATM network using:
ADD !<port> -PORT VirtualPort <path> TRATM
ADD !v2 -PORT VirtualPort 4 TRATM
When only transparent bridging is being used, the source route ring number may be omitted.
To perform Token Ring LAN Emulation client source routing over an ATM network, see Figure 58 and follow these steps:
Figure 58 LAN Emulation Client Source Routing
The following procedure shows how to set up a source routed LEC where both NETBuilder bridge/routers join the same emulated source routed LAN. Notice that both source routed LEC ports must be configured with the same ring number.
On NETBuilder bridge/router A, follow these steps:
1 . Specify the emulated LAN name by entering:
SETDefault !v1 -ATMLE ElanName = "elan3"
2 . Create a virtual port for a Token Ring LEC to be attached to the ATM network using:
ADD !<vport> -PORT VirtualPort <path> TRATM
ADD !v1 -PORT VirtualPort 1 TRATM
3 . Establish the source route ring number 30, using:
SETDefault <!vport> -SR RingNumber=<1...4095>
SETDefault !v1 -SR rn=30
4 . Configure the ring number for the Token Ring port, the bridge number, and turn on the bridge by entering:
SETDefault !1 -SR RingNumber=10
SETDefault -SR BridgeNumber=3
SETDefault -BRIDGE CONTrol=Bridge
On NETBuilder bridge/router B, follow these steps:
1 . Specify the emulated LAN name by entering:
SETDefault !v1 -ATMLE ElanName = "elan3"
2 . Create a virtual port for a source routed LEC to be attached to the ATM network using:
ADD !<vport> -PORT VirtualPort <path> TRATM
ADD !v1 -PORT VirtualPort 1 TRATM
3 . Establish the source route ring number 30, using:
SETDefault <!vport> -SR RingNumber=<1...4095>
SETDefault !v1 -SR rn=30
4 . Configure the ring number for the Token Ring port, the bridge number, and turn on the bridge by entering:
SETDefault !v1 -SR RingNumber=20
SETDefault -SR BridgeNumber=3
SETDefault -BRIDGE CONTrol=Bridge
To verify your ATM LAN emulation configuration, display current ATM configuration information use:
SHowDefault !<port> -ATMLE CONFiguration
Verify that your ATM LAN emulation configuration parameters are configured correctly.
During initialization the LEC can either rely on the ATM switch unit management entity (UME) to determine the ATM address of the LEC's LAN emulation configuration server or configure the ATM address of a specific LECS.
The LECSATMAddr parameter specifies the ATM address of the LECS. When the LEC is in "manual" mode, and the LECSATMAddr parameter is configured, the LEC uses the configured ATM address to connect to the specified LECS. When the LEC is in "automatic" mode, it uses the UME to retrieve the LECS ATM address that will be used during initialization.
To specify which LECS address to use during initialization, follow these steps:
1 . Specify the ATM address of the LECS to be used during initialization using:
SETDefault !<vport> -ATMLE LECSAddr <atm address>
SETDefault !v4 -ATMLE LECSAddr 47007900000000000000000A03E000000100
2 . Set the LEC to manual mode using:
SETDefault !<vport> -ATMLE CONTrol = ([ MANual | AUTOmatic ], [ Proxy | NoProxy ])
SETDefault !v4 -ATMLE CONTrol = MANual
3 . Enable the LEC virtual port by entering:
SETDefault !v4 -POrt CONTrol = ENabled
Multiprotocol over ATM (MPOA) is a means to provide short-cut, inter-subnet virtual circuit connections in a LAN emulation topology. MPOA architecture consists of MPOA clients (MPCs), and MPOA servers (MPS).
Figure 59 is an example of an MPOA configuration.
Figure 59 Multiprotocol over ATM Configuration Example

To configure the MPS routers shown in Figure 59, follow these steps:
1 . On the bridge/router MPS A, specify the emulated LAN names by entering:
SETDefault !v1 -ATMLE ElanName = "elan_1"
SETDefault !v2 -ATMLE ElanName = "elan_2"
2 . Create virtual ports using:
ADD !<port> -PORT VirtualPort <path> ETHATM
ADD !v1 -PORT VirtualPort 1 ETHATM
ADD !v2 -PORT VirtualPort 1 ETHATM
3 . Assign an IP address to each of the virtual ports on MPS A entering:
SETDefault !v1 -IP NETaddr = 10.0.0.1
SETDefault !v2 -IP NETaddr = 20.0.0.1
4 . On the bridge/router MPS B, specify the emulated LAN names by entering:
SETDefault !v1 -ATMLE ElanName = "elan_2"
SETDefault !v2 -ATMLE ElanName = "elan_3"
5 . Create virtual ports using:
ADD !<port> -PORT VirtualPort <path> ETHATM
ADD !v1 -PORT VirtualPort 1 ETHATM
ADD !v2 -PORT VirtualPort 1 ETHATM
6 . Assign an IP address to each of the virtual ports on MPS B by entering:
SETDefault !v1 -IP NETaddr = 20.0.0.3
SETDefault !v2 -IP NETaddr = 30.0.0.3
7 . On the bridge/router MPS C, specify the emulated LAN names by entering:
SETDefault !v1 -ATMLE ElanName = "elan_3"
SETDefault !v2 -ATMLE ElanName = "elan_4"
8 . Create virtual ports using:
ADD !<port> -PORT VirtualPort <path> ETHATM
ADD !v1 -PORT VirtualPort 1 ETHATM
ADD !v2 -PORT VirtualPort 1 ETHATM
9 . Assign an IP address for each of the virtual ports on MPS C by entering:
SETDefault !v1 -IP NETaddr = 30.0.0.4
SETDefault !v2 -IP NETaddr = 40.0.0.4
You may also want to adjust the operational setting for the Multiprotocol Over ATM Server using the MPS Service parameters. For more information, see the MPS Service Parameters chapter in Reference for NETBuilder Family Software.
Normally, the routed path for packets exchanged between subnets must go through routers connecting the subnets. In Figure 59, station A and station C must go through the routers connecting ELAN_1, ELAN_2, ELAN_3, and ELAN_4. Since the edge devices providing station A and station C access to the ATM network are both physically attached to the same ATM network fabric, the edge devices should be able to connect directly with each other, therefore allowing station A and station C to bypass the intermediate routers in the data path. MPOA provides the capability for an edge device to resolve the ATM address of the edge device servicing a destination network protocol address. The edge devices connect to each other and bypass the intermediate routers.
In Figure 59, the packet enters the MPOA system at the incoming MPC (MPC1). MPC1 has already determined that the next hop router's MAC address belongs to a MPS (when the LE_ARP_RESPONSE packet resolved the router's MAC address to ATM address), MPS C. MPC1 creates a cache entry for the destination Internetworking address (e.g. IP address) and begins monitoring the flow to that destination. Once a flow is detected (number of packets sent to that destination exceeding some threshold), MPC1 puts together a MPOA Resolution Request for that destination and sends it on the MPOA VCC to MPS C.
When MPS C receives the MPOA Resolution Request, it examines the destination address specified in the MPOA Resolution Request. The destination address subnet is not a locally attached network. The next-hop towards the destination address is the router, MPS B. MPS A discovers that the MAC address associated with MPS B belongs to another MPS and re-originates the MPOA Resolution Request as a NHRP Resolution request. The packet is forwarded on the routed path through LANE Data Direct VCC to MPS B toward the destination. The re-originated NHRP request will have the MPS C's protocol address as the source protocol address and a new NHRP request ID derived from mapping the source ATM address, destination protocol address, and MPOA request ID.
MPS B receives the NHRP Resolution Request and determines that the next hop MAC address toward the destination specified in the request is another MPS (MPS A) and forwards the request on the routed path through LANE Data Direct VCC to MPS A.
MPS A receives the NHRP Resolution Request and determines the destination address subnet is a locally attached network. MPS A inspects its MPOA cache and discovers that the destination protocol address next hop MAC address belongs to a MPC that it services. MPS A translates the NHRP Resolution Request to an MPOA Cache Imposition Request and sends it on the MPOA VCC to the outgoing MPC (MPC2).
MPC2 receives the MPOA Cache Imposition Request and creates a cache entry and responds to the Cache Imposition Request by returning an MPOA Cache Imposition Reply on the MPOA VCC to the outgoing MPS (MPS A).
The outgoing MPS (MPS A) then translates the MPOA Cache Imposition Reply to an NHRP Resolution Reply and forwards the reply through LANE Data Direct VCC on the routed path toward the incoming MPS (MPS C).
MPS B receives the NHRP Resolution Reply and forwards it through LANE Data Direct VCC toward the incoming MPS (MPS C). When the incoming MPS (MPS C) receives the NHRP Resolution Reply, it matches the ATM address, destination protocol address, request ID with an outstanding NHRP Resolution Request. MPS C translates the Reply to an MPOA Resolution Reply and sends it on the MPOA VCC to the incoming MPC (MPC1).
At the end of this process, the incoming MPC (MPC1) is prepared to establish a MPOA short-cut VCC and the outgoing MPC (MPC2) is prepared to receive data over the short-cut. MPC1 opens a VCC connection to MPC2 and associates the short-cut VCC connection to the IP address that had initiated the MPOA Resolution Request. When MPC1 detects a packet destined to that IP address, it will send the packet over the short-cut VCC connection.
ATM transmits voice, video, and data across LANs, MANs, and WANs. ATM is an international standard defined by the American National Standards Institute (ANSI) and the International Telecommunications Union-Telecommunications Standards Sector (ITU-TSS), formerly CCITT. ATM is the result of research and the development of the Broadband Integrated Services Digital Network (B-ISDN).
ATM implements a high-speed, connection-oriented, cell-switching, and multiplexing technology that provides bandwidth up to 155 Mbps (NETBuilder's offering). In ATM, all information is formatted into small, fixed-length units called cells. Each cell contains 53 octets divided into a 48-octet information field (or payload) and a 5-octet header. By using small fixed-length cells with switching technology, ATM can provide minimal delays for voice and video applications. The switch processes each cell more quickly, and the switch throughput increases. Since packets are broken into small cells that are multiplexed on the ATM network backbone, smaller time-sensitive packets are less likely to be delayed by large packets on the network as often happens in a LAN environment.
ATM operates in a connection-oriented mode. A connection-oriented service requires that a virtual connection be established between the source and destination nodes before data can be transmitted. All connections are virtual in the sense that bandwidth is not permanently assigned to the connection; instead, the network provides the required bandwidth when cells are transmitted. Connections can be established at subscription time as permanent virtual circuits (PVCs) or on demand as switched virtual circuits (SVCs) using a signaling protocol.
The NETBuilder II software supports the ATM Forum's ATM LAN Emulation User Network Specification version 1.0.
The interface for interoperability with legacy LANs and protocols is the LAN emulation user network interface (LUNI) shown in Figure 60. The LUNI protocols allow ATM-attached end systems and LAN/ATM conversion devices to control the virtual connections required for transmission and to emulate the connectionless nature of a LAN or LAN emulation.
Figure 60 LAN Emulation User Network Interface (LUNI)
The main objective of the LAN emulation specification is to enable existing applications to access an ATM network through protocol stacks such as APPN, NetBIOS, and IPX as if they were running over traditional LANs.
LAN emulation works at the media access control (MAC) layer and enables legacy Ethernet, token ring, or FDDI traffic to run over ATM with no modifications to applications network operating systems, or desktop adapters. Legacy end stations can use LAN emulation to connect to other legacy systems as well as to ATM-attached servers, routers, hubs, and other networking devices.
The header of each ATM cell contains addressing information like traditional LAN packets. Instead of a specific destination address, each cell contains two fields, an 8-bit VPI and a 16-bit VCI, that specify the PVC or SVC over which the cell should be forwarded. The VPI and VCI fields define a routing field that provides an ATM switch with the information that it needs to route the cell. The PVC or SVC is usually represented in VPI.VCI format, where VPI is a decimal number between 0 and 255 and VCI is a decimal number between 0 and 65,535.
LAN emulation is a method for carrying network layer packets across an ATM network. The function of the LAN emulation protocol is to emulate LAN while transporting the packets over an ATM network. The LAN emulation protocol defines the service interface for higher layer network protocols. This interface presents an identical appearance to the existing LANs, and data sent across the ATM network is encapsulated in appropriate LAN MAC packet format. The MAC protocol of the specific LAN is not emulated, whether the MAC protocol is either token passing for 802.5 network types or CSMA/CD for Ethernet types.
An emulated LAN on an ATM network consists of the elements shown in Figure 61.
Figure 61 LAN Emulation Entities
The LEC is a process in the NETBuilder II software that operates as an end system. The LEC forwards data, resolves addresses, and performs control functions for a single end-system. A LEC also provides a standard LAN service interface to any higher layer process that interfaces to the LEC.
Each LEC is identified by a unique ATM address, and is associated with one or more MAC addresses reachable through that ATM address.
The LECS is a process that assigns individual LAN emulation clients to particular emulated LANs by directing them to the LES that corresponds to the ELAN. There is logically one LECS per administrative domain, which serves all LAN emulation clients within that domain.
The LES provides the control functions for a particular emulated LAN. There is only one logical LES per emulated LAN, and to belong to a particular emulated LAN means to have a control relationship with that emulated LAN's particular LES. Each LES is identified by an ATM address. The LES ATM address is supplied to the LEC by the LECS or configured through the user interface.
The Broadcast and Unknown Server (BUS) is a multicast server that is used to flood unknown destination address traffic and forward multicast and broadcast traffic to clients within a particular ELAN. Each LEC is associated with only a single BUS per ELAN, but there may be multiple BUSs within a particular ELAN. The BUS to which a LEC connects is identified by a unique ATM address. The BUS ATM address is supplied to the LEC by the LES.
The operation of a LAN emulation system consisting of the components described above consists of three main phases:
When the interface becomes active, the LEC must get its ATM address. The LEC then sets up a configuration-direct connection to the LECS. The LEC must find the location of the LECS. The LECS address may be configured in the LEC and the NETBuilder II bridge/router set to Manual so that the LEC sets up the configuration-direct connection with the specified LECS. The LEC also can rely on the UME of the ATM switch to determine an appropriate LECS address.
After finding the location of the LECS, the LEC establishes a configuration-direct VCC to the LECS. When successfully connected, the LECS uses a configuration protocol to inform the LEC of the information it requires to connect to its target ELAN. This information includes the ATM address of the LES, the type of LAN being emulated, the maximum packet size on the emulated LAN, and the emulated LAN name, which consists of a text string. Network management usually configures the LECS with this information.
When the LEC gets the LES address, it sets up the control-direct VCC to the LES. When this setup is complete, the LES assigns the LEC with a unique LEC Identifier (LECID). The LEC then registers its own MAC and ATM address with the LES.
The LES then sets the control distribute VCC back to the LEC by adding the LEC as a leaf to a point to multipoint connection. The control direct and distribute VCCs can then be used by the LEC for the LAN emulation ARP (LE_ARP) procedure for requesting the ATM address that corresponds to a particular MAC address. To do this, the LEC formulates an LE_ARP request and sends it to the LES. If the LES recognizes this mapping, it may choose to reply directly on the control-direct VCC. If it does not, it forwards the request on the control-distribute VCC to solicit a response from a LEC that knows the requested MAC address.
If a LEC can respond to the LE_ARP request because it is proxying for that address, the LEC responds to the LES on the control direct VCC. The LES then forwards this response either only to the requesting LEC, or, optionally, on the control-distributed VCC to all LECs. All LECs then can learn and cache the particular address mapping, preventing future LE_ARPs for that MAC address.
To complete registration, a LEC uses this LE_ARP mechanism to determine the ATM address of the BUS. The LEC determines the address by sending an LE_ARP for the MAC broadcast address to the LES, which responds with the BUS ATM address. The LEC then sets up the multicast-send VCC to the BUS. The BUS, then sets up the multicast forward VCC back to the LEC by adding the LEC as a leaf to a point-to-multipoint connection. The LEC is now ready to transfer data.
When a LEC is ready to transmit a data frame onto an ELAN, it first checks its local tables to determine if the ATM address associated with the destination MAC address has already been learned. If it has not, the LEC sends the data frame to the BUS, which delivers a copy of the frame to every client on the ELAN
Simultaneously, the LEC sends an LE_ARP request to the LES, trying to resolve the unknown MAC address. The LE_ARP message includes the source ATM address of the LEC making the request. The LES searches its database of MAC address-to-ATM address mappings and returns the ATM address if known through an LE_ARP response. However, in most implementations the LES forwards the LE_ARP to all clients.
The target client recognizes the MAC address and sends an LE_ARP response to the LES, which includes both its own ATM address and the source ATM address for the LEC originating the LE_ARP request. The server forwards the response message with the target ATM address to all the LECs in broadcast fashion. The cycle ends when the originating LEC recognizes its own ATM address contained in the response. At this point, it has learned the ATM address associated with the unknown MAC address and can set up a data-direct connection to the target LEC. When the LEC is ready to transmit subsequent data frames to the newly learned MAC address, the frames are forwarded on the associated data-direct VCC.
Each LEC builds up its own table of MAC addresses, ATM addresses, and VCC bindings. If a particular MAC address has not been active for some time. The LEC eventually drops it from its cache. When there are no more MAC addresses associated with a data-direct VCC, the connection will eventually be released due to inactivity.
An MPOA configuration consists of MPSs, which are co-located with routers and next hop servers, and MPCs, which are co-located with LECs on MPOA hosts or edge devices.
In nonbroadcast, multiaccess networks such as ATM, all nodes are physically capable of communicating to each other via a direct virtual circuit connection (VCC). In the acceptance of ATM, networking customers using ATM for workgroup LANs and LAN backbones require coexistence with existing legacy LAN networks.
ATM Forum's LAN Emulation provided this migration path by allowing end stations (workstation and servers) to connect to the ATM network as though the end stations were connected to a LAN. LAN Emulation provides the ATM services that emulate the services of existing connectionless and multicast capable legacy LANs across a connection-oriented ATM network.
LAN emulation divided the ATM network into multiple Logical Internet Subnets (LISs) or Emulated LANs (ELANs), but required all inter-LIS traffic to go through routers that connected the LISs.
The penalty for this organizational convenience is that all traffic between the subnets must go through the router, rather than straight through the switch fabric. On a large site it is quite likely that there would be two or more routers on the data path between the end stations. If the two end stations are both physically attached to the same ATM network fabric, then the end stations should be able to communicate directly with each other, bypassing one or more intermediate routers in the data path.
The ATM Forum addressed this inefficiency by using the IETF RFC Next Hop Resolution Protocol (NHRP) to allow inter-subnet Internetworking Layer protocols to communicate over short-cut VCCs. NHRP allows intermediate routers connecting NBMA subnetworks to be bypassed on the data path by allowing the source station to resolve the NBMA address of the next hop toward the destination network protocol address.
NHRP consists of Next Hop Clients (NHCs) and Next Hop Servers (NHSs). The NHC (endstation) initiates the NHRP requests to the NHS (routing entity) to resolve the NBMA address of the NHC serving the destination network protocol address. The NHS contains the NBMA addresses of the stations (NHCs) that it serves through registration from the NHC or by configuration.
Using NHRP does not exactly fit in LANE. The NHRP protocol provides the means to find the NBMA, but it also require a network protocol layer at the NHC (end points of the short-cut VCC) to deliver the packet to the source station. In LANE, the NHC functionality would reside in the LANE edge device (such as the CoreBuilder 7200 module in the CoreBuilder 7000 hub), but the LANE edge device is normally a bridge to a LAN and does not necessarily have an internetwork address or a internetworking layer protocol stack.
Since the MPC might not have a network protocol layer to resolve the destination protocol address to the data link layer address, additional MPOA specific control messages were used to augment the NHRP control messages. Using the MPOA control messages (specifically MPOA Cache Imposition Request and Reply), the MPC caches the data link layer information to allow the MPC to perform network layer forwarding, even though the MPC does not have a network protocol stack.
MPOA solves this problem by integrating LANE and NHRP to preserve the benefits of LAN Emulation and using NHRP to resolve the ATM address of the edge device servicing the destination network protocol address.
MPOA provides MPCs and MPSs and extends the NHRP packet type to include specific MPOA packet types for MPC and MPS communication. MPCs issue queries for ATM addresses and receive replies from the MPS using the MPOA packet types. Communication between MPS and MPS are done using the NHRP packet types.
The MPS are logically co-located with routers and performs the MPOA specific functionality (interacting with the MPC). To perform the NHRP functionality (forwarding of NHRP packets to the outgoing MPS), the MPS must have a NHS. The MPS make use of the standard internetworking protocols such as OSPF and RIP.
The MPC are normally co-located with an edge device that does not contain a internetworking layer protocol stack. To act as a internetwork forwarder when a network protocol packet is received on the shortcut VCC, the MPC contains a internetwork layer forwarding database that is administered by the MPS (containing the router).
The learning of a co-located MPC with an edge device or a co-located MPS with a router is automatically done through LANE because the LAN Emulation Clients (LECs) within the edge device and router already communicate with each other through LANE protocols. MPOA requires the extended TLVs defined in LANE Version 2.0 to allow the LECs to advertise their MPOA capability. Therefore LECs supporting LANE Version 1.0 in an MPOA architecture must support the LANE Version 2.0 features for MPOA.
The NETBuilder II bridge/router provides IEEE 802.5 Token Ring LAN Emulation Client functionality. The Token Ring LEC includes all the functionality described above for IEEE 802.3 emulated LANs, as well as the source routing capabilities of IEEE 802.5 LANs. Token Ring LECs are capable of transparent bridging, source route bridging, and routing either with or without source route discovery. Since Token Ring LECs provide virtually identical operation to Ethernet LECs for transparent frames, only the additional functionality associated with source routing is discussed in this section.
Source route LECs require information about the network topology to make forwarding decisions for source routed frames. This information takes the form of ring number, bridge number combinations, called Route Descriptors (RDs). A source route LEC uses RDs with the LE_ARP function to obtain the ATM address of the LEC attached to a particular ring in the source route path. These RD-to-ATM address mappings are stored by the LEC in a table similar to that used for MAC address mappings on IEEE 802.3 ELANs, and are used to setup data-direct VCCs to send traffic to the destination RD.
To allow the LES to respond to LE_ARPs and provide RD mapping information to other LECs attached to the ELAN, a source route bridge LEC registers all of its RDs with the LES when joining the ELAN. The LEC constructs an RD for each source route ring behind the LEC using the ring number together with its bridge number. These RDs are delivered to the LES on the control-direct VCC using the LAN Emulation registration protocol.
There are three basic types of data frames used in source route networks. The first two, All Routes Explorer (ARE) and Spanning Tree Explorer (STE), are broadcast frames. These frames are sent to the BUS to be delivered to all the LECs on the ELAN. On a source route bridge LEC, ARE frames are always forwarded, while STE frames are only forwarded for ports that are in the forwarding state as determined by the spanning tree protocol.
The third basic type of data frame used in source route networks is the Specifically Routed Frame (SRF). The source route LEC handles these frames in one of two ways based upon the information stored in the Routing Information Field (RIF) of the frame. If the source route LEC determines that the frame is destined to a station on the local ring (the attached ELAN), the frame is forwarded based upon the MAC address-to-ATM address mapping in the same way as on IEEE 802.3 ELANs (described earlier).
If the SRF is destined to a station on a ring beyond the local ring, the LEC uses the information in the frame's RIF to construct an RD consisting of the next-hop bridge number and ring number. The LEC then checks to see if this RD has already been entered in its local RD table. If it has not, the LEC sends an LE_ARP request to the LES for the frame's RD and queues the frame until a data-direct VCC is established (a maximum of 20 frames may be queued).
When the LES receives the LE_ARP request for the RD, it looks in its table of registered RDs to find the ATM address of the LEC that registered the specific RD. The LES gives this ATM address to the LEC in an LE_ARP reply message. The LEC then saves this RD-to-ATM address mapping in its local RD table. If the LEC does not already have a data-direct VCC to the ATM address associated with the RD, it sets up the connection. At this point any frames that have been queued up for this data-direct VCC are transmitted, and as long as the RD is active, all subsequent SRFs destined to the RD are forwarded on this connection.
If a particular RD is not used for a period of time, it will be dropped from the RD table. When all the RDs associated with a particular data-direct VCC have been removed from the RD table, the connection is released.
The following terms are used in this chapter to explain ATM: