NSX Architecture availability best practices

[lucasbernadsky]

Hello, we have a 2 site configuration with a vSAN stretched cluster across the 2 sites and 1 local cluster each site.

Each cluster is prepared for NSX and they have a single transport zone

One of our apps can't be placed in the streched cluster for availability reasons.

So all the App's VMs are deployed on the non stretched clusters.

We have the app deployed in site A and replicated to site B by the same app clustering features.

Currently NSX edges are deployed on the  local clusters (one each site) and DLR CVM's are deployed on the stretched cluster (we have one for site A and one for site B)

Do you think this is optimal form a resiliency perspective or this can be improved?

[nachogonzalez · Accepted solution]

Hello, hope you are doing fine

Since you can't deploy all the virtual machines on the stretched cluster (which would've been the easiest configuration)
I'd suggest the following

Create a DLR CVM on  Site A cluster and a DLR CVM for Site B each of those connected to it's ESG.

Please consider enabling NSX HA on both Edge Services Gateway and CVM DLR  and keeping anti affinity rules so both vms are always separated.:

[lucasbernadsky]

Okay, that makes total sense! I was heading that way. Thanks!