Question: What changes in behavior occur when specifying multipoint vs. poin-to-point on a sub interface in Frame Relay. The reason for my question is that I am reviewing someone else's config and I noticed that they are specifying multi-point on the subinterfaces instead of point-to-point as I would expect and I need to be able to explain why this might be a bad thing.
Thanks for the reply. I was looking for a little deeper understanding of the mechanics involved. In other words, what will happen if I build subinterfaces like I am doing traditional point-to-point (mapping each PVC to a separate sub-interface on a separate IP subnet), however I screw up and use the keyword multipoint instead of point-to-point. What do I break (if anything) by doing that?
On a point-to-point subinterface, one PVC will be assigned to the sub-interface, and all packets sent over that sub-interface will be mapped to that PVC. On a point-to-multipoint sub-interface, network address-to-DLCI mapping will be more of an issue, as the router must learn of the mappings via inverse arp or through frame-relay map statements in the configuration. The point-to-multipoint sub-interface will behave just as a physical serial interface with frame-relay encapsulation, with certain DLCIs assigned to the sub-interface.
Please note that when running OSPF over frame relay, all routers must have the same interface type as their neighbors.