Clusering SQL 2012 on ESX 5.1

[aetius1980]

Hi,

Can someone tell me the actual status of Microsoft Cluster Services with Vmware Esxi 5.1? This link says it supports HA and DRS:

VMware KB: Microsoft Clustering on VMware vSphere: Guidelines for supported configurations

This link says it doesn't:

http://pubs.vmware.com/vsphere-51/index.jsp?topic=%2Fcom.vmware.vsphere.mscs.doc%2FGUID-6BD834AE-69B...

I'm looking at using SQL AlwaysOn Availability Groups on an ESXi  5.1 cluster and want to confirm the status of HA and DRS support on this cluster? My SQL AlwaysOn Availability Group will not be using shared storage (file share witness will be used), but it still relies on Microsoft Clustering Services. I'm thinking of creating 2 Windows 2012 Server VMs, setup Windows clustering services with eager zeroed thick VMDKs (not RDMs), on a cluster with HA and DRS enabled, but DRS and HA disabled for my cluster VMs. Is this a supported scenario?

It would be great to hear from people who have something similar.

thanks

[rh5592 · Accepted solution]

Follow the KB article. With your setup which do not rely on shared disk, the SQL 2012 VM's in the AlwaysOn Availaibity group will just be another VM and can be protected by HA. It can also utilize vMotion and DRS.  Just setup anti-affinity rules to ensure that the SQL VM's don't sit on the same ESXi host.

Added info here http://www.vmware.com/files/pdf/solutions/SQL_Server_on_VMware-Availability_and_Recovery_Options.pdf

[NuggetGTR]

HA works and is supported but if SHARED storage is being used then DRS is not supported. but if you using availability groups with out using FCI (Failover Cluster Instance) then there is no shared storage so the normal rules of MSCS do not apply. When people talk MSCS they are generally referring to fail-over clustering using shared storage which limits what can be done.

[aetius1980]

Thanks, so just to recap - I can setup AlwaysON Availability Groups which relies on Windows Fail Over Clustering (I'll be using a file share witness) with DRS if I choose.

I'll most like disable DRS anyway as I've seen issues in production environment with clustering Database Availability Groups and DRS. DRS kicks in, moves an active datbase to a different node and suddenly exchange falls over - not sure how much better ESXi 5.1 is than the older versions, but I'll turn off DRS to avoid scenarios where DRS kicks in, moves a VM and the cluster detects a blip on the network before taking an active node offline