Sql-server – Failover Cluster Migration Wizard (SQL Install Portion)

clusterfailoverclustersql server

Have 5 SQL 2005 to 2008 R2 clustered instances. I've setup a new cluster to prepare to use the cluster migration wizard. As I understand it you have to install SQL before running the cluster migration wizard. My question as it relates to this is, what type of install do I do?

Should I just setup a new standalone SQL install, do the migration wizard, and then uninstall the standalone instance? Or do I need to do a clustered install?

The MS pages as it relates to SQL, don't go into these specifics: http://technet.microsoft.com/en-us/library/ff182337(v=ws.10).aspx

Thanks!

Best Answer

I wouldn't trust the migration wizard to copy everything over correctly for SQL Server. I can see this leaving you with a broken SQL Server instance. What I would recommend doing would be the following...

  1. Build the new cluster with the new OS and SQL Server versions with SQL as a clustered instance with the same instance name (the OS name will be different, but we'll deal with that later).
  2. Copy all the logins, SSIS packages and jobs to the new clustered instance.
  3. On the night of the upgrade take the old clustered instance offline.
  4. Take a SAN snapshot of the LUN (this will be your rollback)
  5. Move the LUNs from the old cluster to the new cluster and bring the LUNs online and add them as clustered resources.
  6. Put the new clustered disks into the SQL Server resource group.
  7. Make the SQL Server service dependent on the clustered disks within the failover cluster manager.
  8. Attach the databases to the new clustered instance.
  9. Add a new network name resource to the cluster based on the old clustered instance's network name (this will probably require that you delete the network name from Active Directory first).
  10. Add a new network IP resource to the cluster based on the old clustered instance's IP address (optional)
  11. Test
  12. Once testing is complete delete the SAN snapshot.
  13. Done

If rollback is needed because something isn't working you detach the databases from the new instance, move the LUNs back to the old cluster, then roll back the SAN snapshot and reattach the databases to the old version of SQL Server.

Related Topic