Network Computingwww.networkcomputing.com

Within each topology, the routing layout among sites can be handled in a variety of ways as well, allowing an administrator to prioritize the Data Link Connection Identifier (DLCI), an address identifying a permanent virtual circuit (PVC) end point to maximize performance over PVCs to each site. For this, a FRAD must have versatile DLCI and routing support. Some FRADs autodetect the DLCIs and their routes from the provider's switch for added simplicity of administration.

Organizations using carriers that support this in their switches can avoid the complexity and cost of deploying and maintaining these rem ote sites by letting the FRAD figure everything out. The advantages are more evident as changes are made to routes and the topology of the network.

By supporting several PVCs simultaneously, frame relay can connect multiple sites through a single physical connection. If the frame relay carrier and the FRAD both support priority PVCs, different traffic types can be assigned to specific DLCIs, which can then be mapped to the priority PVCs to ensure that time-sensitive traffic gets priority over less time-sensitive data types.

This is particularly useful if time-sensitive data, such as SNA and voice traffic, is being transmitted along with less time-sensitive IP, IPX and other traffic. It's also important to be able to consolidate multiple traffic streams into a single DLCI to reduce PVC fees from the frame relay carrier.

Frame relay is capable of supporting both perma nent and switched virtual circuits, and service providers typically charge for each of these paths. FRADs, as well as the frame relay carrier, must support permanent and switched virtual circuits in order to take advantage of the services.

Which CIR, Sir? As for tariffs, organizations are charged at predicted information rate, also known as a committed information rate (CIR). Frame relay providers will guarantee delivery of packets up to this CIR. With this setup, sites that don't require constant high bandwidth pay less money for the service. A carrier's publicly shared network generally will deliver traffic even if it is above CIR for a short time. However, any traffic sent at a rate above the CIR may be discarded, and there's no guarantee those packets will be delivered if the frame relay network is congested. Good FRADs will support the option for burst rates and burst excess data rates, occasionally getting you more bandwidth than you pay for.

Also, the network cannot enforce congestion flow control. It's up to the FRADs to help control the flow of packets when the network gets too congested. Forward Explicit Congest ion Notification (FECN) is a set of bits in a packet that notify a receiving interface device to take precautions to help avoid further network congestion. Backward Explicit Congestion Notification (BECN) notifies a sending device to stop sending frames until the receiving device is ready again. For more details on this topic, see our Network Design Manual chapter on building frame relay networks (www. NetworkComputing.com/netdesign/frame1.html).

Bridging versus Routing The advantages to bridging are that these networks are simple to configure, transparently relay traffic among sites, and can more efficiently use the network-layer address space. Bridging's biggest problem is its inability to control the amount of traffic being sent over a frame relay connection.

Routing FRADS can filter out packets that can cause congestion over the wide area. Security is added with th e ability to filter out packets. For example, routed IP packets can be filtered out based on their source, destination and t ype.

Additionally, many traffic types sent over frame relay are delay-sensitive, such as SNA traffic, and voice, video and fax. This traffic must be received smoothly at the other end to be useful. Therefore, a FRAD that actually "routes" rather than bridges can intelligently deliver packets to the receiving end.