So I'm still stuggling with this, with a 'go live' deadline of this evening...
Current sitatuion:
we have a 6513 at the core of our network, with 2 MPLS networks servicing our branch offices. We have provider A and B routers doing BGP across the MPLS, and EIGRP locally. We redistribute the routes across the protocols. Life is good.
Now, we are replacing 6513's with 2 new Nexi's (what is the plural of Nexus anyway?) I thought it would be a simple fork-lift. Set up a peer-keep alive and a VPC between the 2 nexus units, connect one router to each nexus (each router has only 1 link to the internal network so there is no redundancy in that regard), and configure EIGRp on the nexus.
Not exactly. I learned variance won't work and got kind help on this forum around that.
Now though, I come across an article from Brad Hedlund on routing across the Nexus units that threw me a curve. The article says that a link like we are talking about won't work, in fact, it looks like from that article that our set up won't work and the best we can do is HSRP with static routes?!
That may be true if your routers have multiple links and you are trying to do VPC from each router, one link to each Nexus. That's not what we are trying to do. Meanwhile our SE says we need a layer 3 link (another vlan or SVI) for our eigrp network that we deliberately stop from traversing all other trunk links...but why? what does that do for us? That's going to put a lot of readdressing on us at a time when it's late and we are tired and error prone.
I'm trying to make this as close to 'fork lift' operation as I possibly can, with as little reconfiguring as possible. In less than a month we hope to have a pFr based answer and anyting I do tonight would be undone anyway. Would I be better at this point putting both routers on a single nexus and dealing with the potential risks if that nexus failed? I'm looking for SIMPLE and DEPENDABLE.
