Thanks Harrold, I agree with you on this.

I got one more input regarding this case which is based on observation and would like to share and discuss with you - correct me if i am wrong

This is how cisco implements :-

VRF (e.g. vrf lite)is basically virtual instance of routing table, so if we run ospf in an VRF its basically a differnt instance of OSPF.

so this is an implementation restriction because OSPF needs distuinguished router-id for all the instance running on a same router.

Now take the case of Redback Networks SE800, this is how they implement the Virtual routing concept:-

we can define differnt contexts (Virtual Router contexts) and OSPF process running in the context have nothig to do with OSPF process of other context so they can have same router-id.

Based on this observation can we say that OSPF has defined guidelines about the router-d and cisco is implementing its VRF concept as Virtual routing instance rather complete virtual router so have to be different router-id for each instance ..

rgds

rajat

Rajat,

This is simply an implementation decision to avoid having issues if LSAs from one VRF were inadvertently leaked between separate VRFs.

Regards,

Sorry for the late reply.

My doubts are now pretty much cleared

Thanks Harrold and Sam for this discussion.

rgds

Sam,

ospf also allows you to configure the RID within the VRF context. But an implementation choice has been made not to allow the same RID to exist in different VRFs, just to avoid any issues if LSA are inadvertently being leaked from one VRF to the other.

Regards,

to clarify this, I read OSPF will only support a different ID under each VRF. which has not been the case for BGP until recently.

Is this right ?

Many thanks !

Sam

Sam,

Correct. BGP is a different story as you can have many VRF instances under the same BGP process. For ospf each process represents a separate VRF.

Regards,