BGP/OSPF design question

[ianmckellen]

I have a basic design question. We have few remote sites connected to the two cores, the two cores (6500)  at two different sites are running OSPF between them and to the remote sites. The two 6500 are running eBGP to another partner, and iBGP between each other. We recive routes from the partner via eBGP, then we redistbuite the routes into OSPF to advertise them to remote sites. We also redistbuite OSPF into BGP to advertize our routes to the partner.

Now for 6500-1, we recive the partner routes from eBGP,  from 6500-2 via iBGP and via OSPF (BGP to OSPF redistbution). Under normal operations, the partner routers coming from eBGP will be prefered (Admin distance 20). If the link to the partner fails on 6500-1, the OSPF (admin distance 110) routes coming from 6500-2 will be prefered over the iBGP (admin distance 200).

My question is, is our design is correct? If so, what is the benifits of running iBGP? when will it be used?

We need all the routes to reach the remote sites not just a default route..

Thanks for your help,

Ian

[Jon Marshall]

Ian

So both 6500s are connected to the partner running EBGP and if one 6500 fails it receives the routes to the partner from the other 6500 via IBGP and OSPF ?

The main reasons to use IBGP would be -

1) if the number of routes was too large to redistribute into OSPF ie. no one would redistribute the entire internet routing into an IGP because your internal devices couldn't handle it

and/or

2) if you want to use BGP to influence the path taken.

So in your case if you wanted to ensure one of the 6500s was always used to send traffic to the partner you could use local preference to make sure traffic always went via the primary 6500 even if traffic from remote sites first went to the other 6500.

Note you could also influence which path is taken by manipulating OSPF cost etc. but BGP gives you a lot more flexibility in how you do this.

So based on the information you have given us IBGP is not doing anything here as far as I can see.

Obviously that doesn't mean you should just turn it off without a full understanding of how everything works because there may be more to it and it isn't doing any harm.

Jon

[ianmckellen]

Thank you so much. I still have some confusion,

Now let's assume that I changed the Local pref to 120 on the eBGP routes recived by 6500-2,

Now 6500-1 has the same routes coming from eBGP, iBGP (with LP=120) and OSPF, which route will it choose under normal operation? And which one will it use when eBGP link fails?

According to Cisco web site, BGP best path selection: the LP is more prefered than iBGP/eBGP paths

Thanks again.

[Jon Marshall]

It should choose the OSPF route the logic being -

it receives an EBGP route with a default value of 100 and an IBGP route with a value of 120 and so it uses the IBGP route because of the BGP best path algorithm.

But the IBGP route has a higher AD than the OSPF route so the OSPF route will be installed pointing to the other 6500.

Note the OSPF route in the IP routing table will be an OSPF external.

If the EBGP link fails it makes no difference because it wasn't using that link anyway.

Or did you mean the EBGP link on 6500-2 ?

Jon

[ianmckellen]

Thank you. I meant eBGP link on 6500-1.