• Describe different methods for connecting to a WAN

Using subinterfaces to solve your split-horizon problem is the best way to go. Subinterfaces enable you to break your single, physical interface into multiple, logical interfaces. You still have only a single physical connection to the Frame Relay service provider, however, your router sees it as multiple connections. There are two categories of Frame Relay subinterfaces:
point-to-point and multipoint. Only the point-to-point interface type is designed to fix the split-horizon issue by creating a subinterface for each PVC connection. However, because we’re talking about subinterfaces, it would be a good idea to talk about multipoint as well.

Multipoint Subinterfaces

Forgive me for allowing my opinion to get in the way of a neutral CCNA book, but multipoint interfaces are about as useful as a squirt gun for a scuba diver. This subinterface type enables you to have multiple PVCs terminating under a single, logical subinterface. This type of subinterface is how a physical interface behaves if no subinterfaces are created. It still encounters the
same split-horizon issues, forcing you to disable split-horizon. You might wonder, “How is this

type of subinterface helpful then?” Like I said…squirt gun for a scuba diver. Multipoint subinterfaces can be useful in other, non–Frame Relay deployments, but in Frame Relay they cause more trouble than anything.

Point-to-Point Subinterfaces

This subinterface type enables you to logically design your Frame Relay network as a series of point-to-point connections, no matter how complex your PVC configuration may be. Each PVC circuit is assigned to a single, point-to-point subinterface. This solves the split-horizon issue without much intervention from you because the Cisco router now sees each PVC as its
own interface.

For example, if a routing update is received on Serial 0, split-horizon blocks that update from being sent back out Serial 0. After you configure point-to-point subinterfaces correctly, the router sees an update come in on one of the subinterfaces, such as Serial 0.10. It has no problem sending that update out another subinterface, such as Serial 0.20, because it sees these interfaces as two, distinct ports.

Address Mapping in Frame Relay


  • Configure and verify Frame Relay on Cisco routers

Frame Relay functions at the Data Link layer of the OSI model. It provides services for the WAN just as ethernet provides services for the LAN. Because everything we do today typically relies on IP addresses, Frame Relay needs to have a way of mapping its Data Link layer address (a DLCI) to a Network layer address (typically an IP address). For example, your
router may know that it can reach some location through DLCI 505, but DLCI 505 doesn’t really mean anything to your router. Your router works with IP addresses, not DLCI numbers.

So to allow DLCI 505 to mean something, your router needs to somehow map this to the IP address that DLCI 505 can reach. There are two methods you can use to accomplish this: Inverse ARP and static mappings.

Inverse ARP
Inverse ARP is the router’s automated method to map DLCIs to IP addresses. It works as follows:

  1. You connect your router to the Frame Relay service provider through a serial interface.
  2. The service provider uses LMI to identify your router and send your router a list of DLCIs it can use to reach your remote sites.
  3. Your router sends Inverse ARP messages to each one of these DLCI numbers. This Inverse ARP message carries the simple message, “Hello DLCI! Please send your IP address.”
  4. The remote router receives the message and responds with its IP address.
  5. Your router maps the DLCI number to the IP address it received.

The router sends these Inverse ARP messages to each DLCI number it has received until it has a complete mapping table of DLCI numbers to IP addresses. All along, you as the administrator have done…nothing! This is a completely automated process that makes the Frame Relay setup seamless for small environments. The only drawback to this method is it does not
work for subinterfaces. For Inverse ARP to function properly, you must leave all assigned DLCIs under the physical interface, which causes this interface to become a multipoint interface (if you have multiple DLCI numbers). As you have already seen, multipoint interfaces have problems with split-horizon.

Static Mappings

The alternative to allowing Inverse ARP to automatically configure your environment is to use static maps. This method is also specific to multipoint-style interfaces. Using this method, you can manually enter the DLCI to IP address mapping for each PVC. This gives you complete control over the mapping process and enables you to have more than one interface (unlike Inverse ARP). The specifics of this configuration, along with the considerations for point-topoint interfaces, are covered later in this chapter.

Know what the differences are between Inverse ARP Frame Relay configuration and static mappings.

About the author


Leave a Comment