Full redundancy on N5K and N3K design

[julien_lau]

Hello,

We deployed a N5k and N3k design.

.But now we want that both N3K cluster can discuss each other even if we loose the N5K layer.

So my question is regarding the NEW VPC link on the diagram attached.

I am a bit scared to create a loop in this design, can you tell how spanning will react or what i have to take care of ?

[Mark Malone]

Hi

not sure if that would be recommended what your doing , you already have plenty of resilience there as your access layer should look  like screenshot above and be dual linked  , never seen a DC setup like what your trying to achieve so i cant predict whats going to happen in a failed scenario , there is a VPC best practice guide you should follow , if thats not in there i wouldn't implement it , Cisco will not support unsupported design methods if something goes seriously wrong

https://www.cisco.com/c/dam/en/us/td/docs/switches/datacenter/sw/design/vpc_design/vpc_best_practices_design_guide.pdf

[Giuseppe Larosa]

Hello Julien,

the new vPC would be used only if a whole vPC towards the N5k fails.

This means a double link failure or the the whole N5k device pair to fail.

Be aware that on N3k you should avoid to use commands like

spanning-tree loop guard default.

I would recommend the use of Bridge assurance that makes STP BPDUs exchange bidirectional like a keepalive (even from ports in Alternate or Backup state)

Hope to help

Giuseppe