An AHV-hosted three-tier application is protected to a DR site with Nutanix Disaster Recovery. During failover, the database tier must start first, then the application tier, then the web tier. The admin needs a Prism Central control that enforces this sequence without manual VM-by-VM power actions. Which control should be used?

In Nutanix Disaster Recovery (formerly Leap), a recovery plan is the runbook Prism Central executes for failover, planned failover and test failover. Its power-on sequence is built from stages: each stage holds VMs or categories, the plan powers on stage 1, waits for the configured delay, then moves to stage 2, and so on. Putting the database VMs in the first stage, the application VMs in the second and the web VMs in the third gives the required order without anyone powering VMs on by hand, and the plan can also remap networks and run in-guest scripts. A protection policy controls how often recovery points are created and where they are replicated; replicating one tier more often changes its RPO, not when it boots. A storage policy with a QoS throttle shapes I/O rates and has nothing to do with power-on order. Anti-affinity rules decide which hosts VMs may share, not the sequence in which they start. Exam caveat: AHV has no general startup-order setting for host or cluster restarts; ordered power-on is a recovery plan feature, and stages are the unit of order. Operational check: run a test failover to an isolated network and confirm in the recovery plan report that the database stage completed before the application and web stages started.