With the increased release cadence (that’s 1 shot in the WS2012 R2 drinking game) that Microsoft has adopted, they want to make it easier for us to upgrade our hosts. The first critical steps in enabling that were actually delivered in WS2012 Hyper-V:
- VHDX: You’re a sucker if you’re still deploying passthrough disks. You’re being negligent, in my opinion, if you’re a so-called-expert (like a consultant) and still typing your customers’ VMs to specific hardware devices. I won’t hold any punches on this, and I don’t have time for defensive excuses.
- Shared-Nothing Live Migration: We can move virtual machine files and the VMs themselves between any mix of clustered and non-clustered hosts.
That latter one is important, because we still cannot do an in-place upgrade of a Failover Cluster. Yes, Microsoft has heard the feedback. Just give them more of it if you are talking to them directly, especially if it’s a local/visiting rep from the Windows Server & System Center product group (giving feedback to a person from the local subsidiary is pointless).
We can use cross-platform Live Migration from WS2012 Hyper-V to WS2012 R2 Hyper-V to get zero-downtime “upgrades”. Scenarios include:
- Doing a Shared-Nothing Live Migration from a WS2012 Hyper-V cluster to a Hyper-V Server 2012 R2 cluster
- Performing a Live Migration from a non-clustered Hyper-V Server 2012 host to a non-clustered WS2012 R2 Hyper-V host where they share common SMB 3.0 (WS2012/WS2012 R2) storage.
- And more!
“Upgrading” a Hyper-V cluster using Cross-Version Live Migration
All mixes of Hyper-V Server and Windows Server 2012 and 2012 R2 are included. This is a one-way Live Migration, from 2012 to 2012 R2. And it means you can move your VMs from 2012 to 2012 R2 without impacting uptime for your services and business operations.
This also means that those of you planning WS2012 upgrades/installs shouldn’t stop. You, of course, should be buying Software Assurance for your VM Windows Server licensing (which includes the host), and can use this zero downtime to get from a great hypervisor to an even better one with minimal effort.