You still get hardware redundancy by clustering the hypervisors together, so you could start your VM workload up on a healthy node.