SQL Server or Normal VM behavior for HA

In the event your organization is considering maintaining a SQL server on a Stretched Management cluster (Metro Cluster) or normal cluster please keep in mind the following:

  1. Normal Operations
  2. Host fails and dedicated resources of vCPU, RAM, are now unavailable
  3. Admission Control triggers an HA event 
  4. Virtual Machine is power cycled-down
  5. DRS migrates Virtual Machine to surviving host in the cluster
  6. Virtual Machine is powered-on based on power policy 

Step 4, 5, 6 would cause an outage unless the SQL DB is clustered. Stretched storage only
will help for migration between hosts but dedicated compute resources to VMs will still cause an outage. 

Even placing a stretched storage solution in a cluster doesn’t magically maintain a fully powered on VM. If you steal the compute resources from anything it’s going to have a bad time. The only way around this would be to setup VMware’s FT where everything is mirrored between Host A and Host B.

Leave a Reply

Your email address will not be published. Required fields are marked *

This site uses Akismet to reduce spam. Learn how your comment data is processed.