My company is going to migrate some services from one on-prem site to another but the database servers will remain in the same location (for now).
I've been asked for "suspend" all the activity of the three AOG cluster nodes (all secondary instances are readable without routing), during a long maintenance window.
Two main options are coming to my mind:
- Maybe not the fanciest. Suspend data movement and then stop SQL Engine service at the secondary replicas first and then stop the primary at the end. When the time to start services comes, start the primary first and the secondaries later in inverse order and resume data movement at the end.
- Disable jobs and intended logins at all instances to enable them again later keeping all AOG activity running but without user access.
I know the "best option" is completely relative to the context, but I would like to read some opinions.