Cisco Support Community
Showing results for 
Search instead for 
Did you mean: 

Welcome to Cisco Support Community. We would love to have your feedback.

For an introduction to the new site, click here. If you'd prefer to explore, try our test area to get started. And see here for current known issues.

New Member

FR sub interfaces

i ahve th e concept of FR subinterfaces and i already know that it was first developed to rectify the routing loop problem (split horizan issue) now plz tell me is my nderstanding right or wrong ??? secondly if we are using classless routing protocol we dont need that......thirdly plz tell me one hting that if i split my interfaces into subintyerfaces how much is the max....and do i split them cuz i have limitation on harware on my router??? last thing is how much bandwitdth they can handle.....??? (subinterfaces)


Re: FR sub interfaces

FR subinterfaces have been implemented to avoid the split horizon rules - you are right.

Classless routing protocol is different - it means the routing protocol support VLMS - masks are not standard classes (/8,/16,/24) - so these protocols can carry mask information also instead of classfull protocols. Classless protocol can support masks like: /18 /25 and so on...

The capacity is of your subinterfaces are equal with your main (physical interface capacity - it's trivial). If you have one FastEthernet interface with 2 subinterfaces, the cumulative capoacity can't hiher than the physical interface capacity.

I don't know any limitation about subinterfaces...



Please rate me if I helped.

Re: FR sub interfaces


as with regard to the number of subinterfaces, that depends on the IOS version and platform. If you issue the exec command 'show idb' it will tell you the number of IDB's (Interface Descriptor Blocks) available. Basically, a subinterface consumes one IDB.

Check this link for additional info on the IDB concept:

Maximum Number of Interfaces and Subinterfaces for Cisco IOS Platforms: IDB Limits

In addition to the other post, subinterfaces in Frame Relay also have the benefit of not having to dedicate a complete physical interface for each DLCI. Imagine having a partially meshed network with 9 spokes and 1 hub, if it were not for subinterfaces, you would need a physical interface for each connection.



Hall of Fame Super Silver

Re: FR sub interfaces

I think that saying that Frame Relay subinterfaces were developed to solve the issue with split horizon is too narrow a reason for the development of subinterfaces in Frame Relay. Split Horizon is only a part of why subinterfaces were developed. Additional reasons for the development of subinterfaces for Frame Relay include the ability to change the behavior relative to multi-access interfaces. A Frame Relay interface is inherently treated as multi-access since it can support multiple DLCIs and can connect to multiple remote routers. The subinterface gives the ability to act as point to point as well as multi-point. The subinterface can change the behavior of requiring frame relay maps to resolve the IP addressing of remote neighbors. Subinterfaces can also change the behavior about forwarding broadcast packets. Subinterfaces also give the ability to specify one security policy (an access list) on one subinterface and a different policy (another access list) on another subinterface. So there are multiple reasons for the development of subinterfaces in Frame Relay.

GNT's explanation of the limit of number of subinterfaces being based on the IDB limit, and that the IDB limit is different for various platforms and for different versions of IOS is quite correct.

The question of capacity of a subinteface is answered by the fact that the subinterface has capacity up to the capacity of the physical interface, limited by the fact that if there is more than one subinterface then the aggregate traffic of all subinterfaces can not exceed the capacity of the physical interface. The other limit on the capacity of the subinterface is the traffic rate specified in the contract with the Frame Relay provider (normal rate and burst rate) and the provider may limit the amount of traffic that it allows on the DLCI.