Token Ring Internetworking

team lib

While most of the internetworking frenzy has been directed at internetworking Ethernet LANs, connecting Token Ring LANs encompasses a separate set of issues. Unlike Ethernet, bridging predominates in Token Ring environments. Source routing, which is used in Token Ring, is a more sophisticated form of internetworking than transparent bridging, and this has staved off the onslaught of routing. Yet, as MIS builds increasingly complex networks, routing will become a more critical element in Token Ring internetworks. When internetworking Token Rings, you have three options: source-routing bridging, routing, and source routing transparent.

Source Routing

The simplest way to interconnect Token Ring networks is to bridge them using source routing. In transparent bridging, which is used for Ethernet LANs, the internetworking takes place at the Medium Access Control (MAC) sublayer of the OSI Data-link layer. In source routing, the internetworking occurs at the Logical Link Control (LLC) sublayer, a half-layer "higher" than the MAC sub-layer. This higher level of operation, combined with a richer frame format, gives source routing a higher level of functionality than transparent bridging.

Unlike transparent bridging, in source routing, the sending and receiving devices help determine the route the frame should traverse through the internetwork. The route is discovered through broadcast frames sent between the source and destination frames .

Source-routing bridge hardware is simpler than routers, making them easier to build and less expensive. Source routing also has advantages over transparent bridging. Because the same path is used for the duration of the session, the process is efficient, except for the setup time to determine the route. Source routing optimally uses parallel and redundant paths, so links are not left idle. It also allows IBM controller timing to be properly set for extended LANs.

Source routing has its disadvantages. The route discovery process requires more overhead than transparent bridging. If many devices are simultaneously performing route discovery, the network slows down considerably. The bridge selects the best path at the time of the session's commencement; if the link is heavily loaded, the two devices are stuck with a possibly inefficient path for the session's duration. Transparent bridges choose a new path per frame. Source-routing functionality must be engineered into the bridges, making them more expensive and complex than transparent bridges.

Two types of route discovery frames are used: All-Routes Broadcast (ARB) and Single-Route Broadcast (SRB). An ARB frame traverses all possible routes between end stations, while an SRB takes only one route between end stations. To determine the best path between sending and receiving stations , the sending station transmits an SRB.

When the SRB reaches its destination, that station issues an ARB. Multiple copies of the ARB travel over all possible routes back to the originating station. Note that an ARB can never travel the same ring twice nor exceed the hop count limitsource routing uses the spanning tree structure, which can be configured so only one copy of the frame reaches each ring.

As the frame passes through a bridge, the bridge inserts its own bridge number and the numbers of its attached LAN segments into the frame's Routing Information Field. Each Token Ring segment and bridge are assigned identifying numbers, and the combination of LAN and bridge numbers is unique for each bridge. The bridge also indicates the maximum frame size, which will be used to negotiate the frame size for 4Mbps and 16Mbps Token Ring.

The originating station receives an ARB frame back for each possible path from the sender to the destination as well as the maximum frame size. The originating station selects the most efficient path based on the first frame to return, the number of hops between the source and destination, and the frame size. For the transmission session's duration, that route is used.

When a frame already contains a route, the frame's Routing Information bit is set to notify the bridge that routing information is available. The source-routing bridge examines this bit to filter frames. If the bit is not set, the bridge ignores it. If the frame is not a broadcast (and therefore is data) and the bit is set, then the bridge checks to see if its own ring and bridge numbers are contained. If not, the bridge ignores it. If a match occurs, the frame is forwarded over the link to the next LAN. The bridge sets bits to indicate that it has been copied , which prevents frames from circulating endlessly in the network.

The majority of Token Ring networks use source routing, but it is not appropriate in some cases. For example, if you need to connect Token Ring and Ethernet networks, bridging won't be sufficient.

Network-Layer Routing

You can also interconnect multiple Token Ring networks with routers. Whereas bridges operate at the Data-link layer, routers operate at the Network layer, one level higher. Routers deliver a higher level of functionality than bridges, but they incur greater overhead. Routers must be built specifically to handle Network-layer protocols, such as TCP/IP or IPX/SPX. Many also include bridging functionality for protocols that can't be routed.

As MIS departments consider consolidating SNA and PC LAN networks into one "super-network," many are looking to routing. Companies with Token Ring LANs also look to routing to provide the same level of security, flow control, and path control that Ethernet LANs and routers provide.

In a routed network, routers communicate with each other, learn where the end-station devices reside, and manage the traffic flow. They require a specific protocol to do so. For example, most TCP/IP routers use the Routing Information Protocol (RIP), but some use Open Shortest Path First (OSPF). Routers can detect congestion on a particular link and send network traffic over a less congested link.

If a particular link fails, the router reconfigures the network around the failure. This convergence presents a strong advantage over bridging. Source routing determines its paths statically; if a link fails, the session must be restarted. With routing, if this convergence happens quickly enough, the user won't notice the difference. If the network does not converge quickly enough, the session is dropped and the user loses the connection.

One problem with routers in Token Ring LANs is most IBM networks don't include routable protocols. SNA/SDLC and NetBIOS are not routable protocols. They both lack Network layers and thus can't be handled by routers.

One solution is to use a router that encapsulates Token Ring traffic into TCP/IP packets. This method adds the overhead of encapsulating your traffic into TCP/IP at the transmission side and de-encapsulating it at the receiving end.

If you encapsulate SNA traffic, you may run into timing issues. In an SNA network, the session will be dropped if the controller does not respond within the preset time. Some MIS shops reconfigure their VTAM tables to allow for a slower response time. Some manufacturers add controller-like functionality into their routers to fool the front-end processor into thinking that the router is a cluster controller. "Poll-spoofing" is a tricky feat to accomplish. Test this solution carefully before you implement it in a production network.

Another solution is to add TCP/IP to your network. If TCP/IP runs on all hosts and LAN servers, TCP/IP routers will manage the traffic. This solution makes the most sense if you are largely a TCP/IP shop or if your MIS department is willing to invest in learning TCP/IP technology.

A third solution is to implement a nonstandard routing method (available from different manufacturers) that gives good response time and can quickly reconfigure a network after a failure.

Source Routing Transparent

Source routing transparent is a truce between the Token Ring source-routing camp and the Ethernet transparent-bridging camp. It provides a way for Token Ring and Ethernet LANs to interoperate .

Source routing cannot operate in a transparently bridged LAN, since the frames lack routing information and the bridge has no way of knowing that the frame should be forwarded. You can use transparent bridges in a source-routing environment; however, the routing information is ignored, and the advantages and overhead of source routing are wasted .

Frame size is a problem when you have both 4Mbps and 16Mbps rings, since transparent bridges can't indicate the correct frame sizes, while source-routing bridges can. In many instances, duplicate transparent and source-routing bridges must be used to carry the two types of traffic, which clearly is inefficient.

The upcoming IEEE Source Routing Transparent (SRT) specification addresses the coexistence and interoperability of Ethernet, Token Ring, and FDDI. A source routing transparent bridge allows both source routing and transparent data to be passed. An SRT bridge will source route frames with embedded routing information and transparently bridge those that lack this information.

Source routing transparent implements two logical paths for the two frame types. All frames are filtered to determine if the routing information bit is set. If it isn't, the frame takes the transparent path. If it's set, then it takes the source-routing path.

Source routing transparent defines three frames that can be used for route discovery: All-Route Explorer (ARE), Specifically Routed Frame (SRF), and Transparent Spanning Frame (TSF). The ARE frame is equivalent to source routing's ARB frame; it traverses all routes between end stations. The SRF is issued in response to the ARE. The TSF lacks routing information but performs the SRB's function.

These new route discovery methods are not supported by all IBM source routing software, so changes in some end-station source-routing software will have to occur.

The SRT standard allows the end station to discover the route, which source routing does not permit. The destination station performs this function in three instances: when it becomes the source, when it gets routes from the ARE sent by the source, or when the destination picks up the route from the data frames that were sent from the source. The latter allows the destination to discover the route without creating overhead.

In SRT, end stations have more robust options for choosing the best route for a particular session. For example, they can reserve more than one route to be used as backup if the primary route fails during transmission. Route selection criteria include the first frame returned, the lowest number of hops, the first route returned with the largest frame size, or any combination of the above. Once the route is determined, the path is inserted in the data frame, and the bridge decides whether or not to forward it based on the routing information.

SRT does not enable a transparently bridged LAN to talk to a source-routed LAN; it cannot translate among different frame formats. SRT allows the mixing of source-routed and transparently bridged networks on the same internetwork, alleviating the need to purchase duplicate hardware.

This tutorial, number 48, by Patricia Schnaidt, was originally published in the July 1992 issue of LAN Magazine/Network Magazine.

 
team lib


Network Tutorial
Lan Tutorial With Glossary of Terms: A Complete Introduction to Local Area Networks (Lan Networking Library)
ISBN: 0879303794
EAN: 2147483647
Year: 2003
Pages: 193

flylib.com © 2008-2017.
If you may any questions please contact us: flylib@qtcs.net