Redundant Link Nexus Cores

[joshwa]

I currently have a network that connect the data center 01 to HQ 01 using 1 link with the the IPADDR scope of 10.10.254.0/24. I typically would think that this should be a point to point link /30 but that is why I am asking for a discussion. What is the best way to set up 2 links with redundancy from DC to HQ. DC is off perm and using lit fiber for both connections. I am running EIGRP currently, but would rather have static routes if possible. In the future the company is looking to move to OSPF.

DC = N5K

HQ = N9K

[Jason Leschnik]

Howdy!

As long as you don't need L2 stretched VLANs then:

1. You could just go with 2 separate-routed L3 links with a /30 or /31 on each link
   1. Drawback here is that if a link goes down you have to wait for a re-convergence
   2. BFD could help with this, but your IGP could possibly need a recalc
2. Another option if you have vPC on both sides (make sure both sites have a different vPC Domain ID and running "layer3 peer-router" on all switches) is to bundle both interfaces into an L2 port-channel with vPC and trunk a transit only VLAN between them only used for the L3
   1. You would have 4 SVIs in that transit VLAN on n5k1, n5k2, n9k1, n9k2
   2. The benefit here is that if a link goes down the IGP shouldn't notice a difference because it's a logical port-channel
   3. At the end of this you would have 3 adj. per switch over the transit DCI transit VLAN

Just a thought

Regards,

Jason.