I just performed this migration this week for our SQL database. I used QSM (Because I was migrating from 32-bit to 64-bit).
- Create new igroups on new filer and verify settings before the outage.
- Snapmirror volumes. Make sure to allow this to complete ahead of time to reduce the length of the outage
- Shutdown the Cluster
- Unmap the LUNs
- Update Snapmirror relationship
- Quiesce and Break the relationship
- Map LUNs to igroups on new filer (I used the same LUN IDs to be safe)
- Rescanned for disks on the hosts
- Brought the disks online for the host that last held the reservations. The disks will show "Reserved"
- Brought the disks online for the other node's in the cluster.
- Startup the cluster and verify the disks show up as they should.
I then failed over all of the cluster resources to the other node to verify they came up properly. I had to do this from the command line (Run as administrator) and use "disk group /status" to find the resources then use "disk group <GROUP> /move:<NODE>
This took about half an hour from shutdown to verify as I had most of the commands on the filer scripted out ahead of time.