What should you do to ensure the highest level of availability for virtual machines (VMs) that run on the cluster?