A. The first version of Hyper-V, which will be released within 180 days of the Release to Manufacturing (RTM) of Windows Server 2008 and is currently functional in the beta version, allows a virtual machine (VM) to be made a fail-over clustered service, which means a VM can move between nodes in a failover cluster, aswith other types of failover services such as a printer or file share. It's not a seamless move with no downtime. The actual process when a failover of a VM is performed is as follows:
- The state of the VM is saved to the shared storage, which means the memory of the VM is saved to a file.
- Once the state save is complete, the clustered disk is switched to another node in the cluster and brought online.
- The new owner of the resource enables the VM and reads the state from the memory dump file bringing the VM online and available.
Because of the service interruption while the state of the server is saved to disk, the disk is failed over to the other server. Then the memory is read from disk and the VM started. This is not an ideal high-availability solution and is better used as a controlled way to move VMs between servers for maintenance and balancing purposes. However, if a server does crash, the VM will be failed over to another node but the previous state of the VM will be unknown so when the VM starts on the new node the startup will be a server crash recovery startup.