point-to-point vs. multipoint in frame-relay

The comment’s on this i found have not been very clearly. I figured out that it might be no really difference between these two configurations.

There is a hard limit in the number of subinterfaces at about 300.

But generally it seems to be a better choice to use point-to-point subinterfaces.

On a point-to-point subinterface, you have to use the frame-relay interface-dlci command to associate a DLCI with the subinterface. There’s no inverse ARP – any IP address that matches the address/subnet-mask is assumed
to be at the end of the DLCI.
On a point-to-multipoint you can either use the frame-relay interface-dlci command and let the inverse-arp generate the frame-relay map, or you can use
the frame-relay map command.
Remember, the frame switch doesn’t know about sub-interfaces, only about a collection of DLCIs. It communicates active DLCI numbers via the ILMI
protocol to the router. The router then has to know which DLCIs are assigned to which subinterfaces. If you haven’t specified which subinterface a DLCI belongs to with either a „frame-relay interface-dlci“ or a „frame-relay map“ then the DLCI is put under the main interface by default.
The inverse ARP protocol is end-to-end between the routers – it has nothing to do with the switches in between.

The difference is also, that it is not possible to configure more than one dlci on a point-to-point interface. The error message is:

%PVC already assigned to sub-interface Serial0/0.3
So if you want to configure more than one dlci on an subinterface, you have to use multipoint. But also pysical interface are multipoint by default.

Leave a Reply