[c-nsp] Network Design - OSPF Convergence Issue
Bruce Pinsky
bep at whack.org
Mon Aug 21 18:13:06 EDT 2006
-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1
Shakeel Ahmad wrote:
> Guys,
>
> we came across a live scenario here and i am not a big OSPF guru...please
> guide me here
>
> + ++ HQ ++
> |
> | -- P1
> | | / / | \ \
> | OSPF
> | | \ \ | / /
> | -- 1 2 3 4 5
> |
> + ++Core++
>
>
> Client is running simple GRE tunnels via 5 different providers (1,2,3,4,5)
> to their headquarter's ISP (P1)... All providers are connected on
> different boundry routers and HQ on a single router..
>
> Now Loopback IP's are advertised via OSPF (single area) for all boundry
> routers of client and HQ....Issue currently faced is: loopback IP's are
> learnt from all GRE tunnels...while changing Costs on different GRE tunnels
> doesn't change the traffic path and CORE learns the path from (low cost GRE
> tunnel boundry router)..while BGP ssession is being used for distributing
> ....
>
> The basic need is to converge the network in between boundry routers ON
> DEMAND and sometimes on incident based events.
>
> Can someone share..what would be the best design...covering ON DEMAND path
> selection via OSPF in this design.
>
> (i guess i confused many things..let me have your questions)
>
Maybe I'm missing something, but if each of the boundary routers has a
tunnel to HQ, how is the HQ router learning all the loopbacks over all the
tunnel interfaces? That would seem to imply that all the boundary routers
are sharing a common subnet somehow or that they are fully-meshed. Is that
the case? Seems like we need to know more about the topology in this case
cause something doesn't seem right.
- --
=========
bep
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.4 (MingW32)
Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org
iD8DBQFE6i/yE1XcgMgrtyYRAk8zAJ9E626ZBPprBMpJskSlF39gG+7LHQCbBCq1
CKvC4QuPf6niess1wK+VcXA=
=fX+l
-----END PGP SIGNATURE-----
More information about the cisco-nsp
mailing list