• To patch (MS-SQL)

    You failover or move the resources to other node, let's call it B. You run the service pack or whatever on Node A. At that point, A is fully patched.

    Now, you fail over to Node "A" which becomes active. And now you patch Node B.

    Now both nodes are fully patched.

    This was a major change from MS-SQL 2005 and MS-SQL2000, where you patch the virtual instance and the patch is being replicated over the network to other node. This caused some issues, because if network fails or something breaks, one of the nodes won't end fully patched ending in a non consistent state or different dll versions. Trust me, it was a pain when failed.