I’ve got the go-ahead to deploy our Windows Server 2008 R2 Hyper-V cluster next week. The plan is pretty simple. New hardware is required for expansion. I’ll be setting that up with Windows Server 2008 R2 instead of 2008. It will be set up as our new cluster, as is required for the “upgrade” which is really a migration. I’ll then use VMM to move the VM’s from the old 2008 cluster to the new 2008 R2 cluster. That won’t be able to take advantage of Quick Storage Migration because the old cluster is still W2008. It’ll require getting maintenance windows for he affected VM’s, shutting them down and doing a cold migration from the old cluster to the new. Once that is done I can break up the old cluster, one node at a time as they are made available. As I get each node I can rebuild it with Windows Server 2008 R2.
Using VMM for the procedure simplifies things. The job optimises the export/import process and it uses BITS to transfer the files safely. The job then mounts the VM, adds the integration components and starts up the VM for you if you’ve elected to do that.
We’ve already migrated to W2008 R2 for whatever un-clustered hosts we run. Once we complete the “upgrade” we’ll be able to use quick storage migration for future operations like this. We’ll also get better performance (it was already quite good) and lower power consumption. I’m quite happy so far. I have one DL380 G5 that is moderately specified and we’ll be getting a 14-1 compression ratio out of it without any sacrifices. We’re already at 10-1 and only using between 30-45% of CPU resources. That is quite cool! When we start going with the G6 series of servers from HP we’ll get even more from each host – all thanks to the power and efficiencies of Windows Server 2008 R2 and Hyper-V.
*OK; where’s my commission?* 😉