vSAN Stretched Cluster and HCI Mesh

[imjoshua]

Hi,

Anyone has experience with vSAN Stretched cluster with HCI Mesh Compute nodes. Is it recommended for production workloads?

The following solution is being considered:

• The total 12 nodes across DC and DR
• 6 node vsan stretched cluster, 3 in DC and 3 in DR , 1 witness in 3rd site
• 6 node HCI Mesh Compute cluster, 3 in DC and 3 in DR

Our Assumptions,

If DC fails,

• All the VMs in vSAN stretched cluster will be restarted to DR HCI Nodes
• All the VMS in the HCI Mesh Compute cluster will be restarted to DR Compute Nodes

Please let us know your thoughts / comments on the architecture

[kastlr]

Hi,

your stretched HCI Mesh cluster would consume storage provided by your stretched vSAN cluster.

So if one side/DC of your stretched cluster went down/offline only the storage (and vSAN objects) which reside on the surviving side would be accessable.

If you properly configure the SPBMs any VM running on nodes located at the surviving side wouldn't be impacted.

All other VMs would crash (due to total outage including CPU, RAM, network & storage) and should be restarted on the surviving side.

May I ask why you created such a small stretched cluster which should than also act as HCI Mesh provider?
With 3 nodes per side you're limited to MIR only, with 4 nodes (or more) you could use EC (if running AFA nodes).