We just tackled this one in our organisation. 2 Gotchas.
1. Router-id must be different between peers, make sure your code supports vrf specific router-id. 2. iBGP was very messy IMHO, so we went with eBGP using local-as to have each vrf appear to be a different 65xxx AS I can sent you my lab config's tomorrow. Thanks, Aaron > -----Original Message----- > From: [email protected] [mailto:cisco-nsp- > [email protected]] On Behalf Of Tim Durack > Sent: Thursday, 18 December 2008 1:54 AM > To: [email protected] > Subject: [c-nsp] MPLS-VPN migration > > Looking for some "creative" ideas on how best to accomplish this: > > We are migrating a traditional enterprise-style IP network to an > MPLS-VPN network. All the infrastructure MPLS/IGP/MP-BGP work is > essentially done (it's a purely PE-PE network, no P routers anywhere.) > > All "customer" networks are still in the global table. I need to > migrate them into VPN groups, but maintain full reachability between > global and VRFs during the migration. Route-leaking will be configured > between VRFs, and at a later stage some kind of firewall will be > employed between VPNs. The hard part is getting everything into the > VPNs first (without anyone noticing too much :-) > > Ideally I'd like to bring up BGP sessions between the global table and > VRFs on each PE. I notice I can do BGP sessions between VRFs, but > can't quite wrap my head around global->VRF BGP. Is this even > possible? > > Thanks for thinking about it. > > Tim:> > _______________________________________________ > cisco-nsp mailing list [email protected] > https://puck.nether.net/mailman/listinfo/cisco-nsp > archive at http://puck.nether.net/pipermail/cisco-nsp/ _______________________________________________ cisco-nsp mailing list [email protected] https://puck.nether.net/mailman/listinfo/cisco-nsp archive at http://puck.nether.net/pipermail/cisco-nsp/
