[c-nsp] BGP questions

[Dan Benson]

I agree with Troy, you should home the route orignation to the 6500's 
and run OSPF between the routers and the switches, this way in the event 
of a failure anywhere within the network the IBGP sessions should be 
self healing, therefore routes between all upstreams should stay 
consistant in the 6500's.  It should be a classic ring typoligy, running 
OSPF and IBGP, peering between all the local (internal AS) devices 
should be done by Loopbacks that are injected into OSPF.  Good luck.. //db