An administrator is tasked with verifying the cluster's disaster recovery capability. The administrator needs to validate the capability without impacting the production VMs on the Source cluster and ensure that the destination cluster can bring up all the protected VMs. How can the administrator perform a failover?

A test failover is the specific feature designed to validate a disaster recovery plan without any impact on the production environment. When a test failover is initiated, the recovery plan brings up the protected VMs on the destination cluster using the replicated snapshots. These test VMs are created in an isolated network bubble, preventing any conflict with the production VMs, which continue to run uninterrupted at the source cluster. This allows the administrator to perform a full validation of the DR process, including VM boot-up and application functionality, in a safe, non-disruptive manner.

A. Validating a Recovery Plan is only a configuration check. It verifies syntax and entity existence but does not power on any VMs or test the actual recovery process.
B. An unplanned failover is a real disaster recovery invocation, used when the source site is offline. It is disruptive and intended for actual emergencies, not for testing.
D. A planned failover is a controlled migration from the source to the destination site. It involves gracefully shutting down production VMs, making it a disruptive, production-impacting event.

1. Nutanix Leap Administration Guide (dr-2023.3), "Recovery Plan Workflows", Section: "Testing a Recovery Plan". The documentation explicitly states: "You can test a recovery plan without impacting your production environment... The test VMs power-on in the selected test network at the recovery AZ."

2. Nutanix University, "Nutanix Data Protection and Disaster Recovery (NDPDR) 6.5", Module 4: "Disaster Recovery with Prism Central". The courseware describes test failover as the primary method for non-disruptive DR drills, contrasting it with the production-impacting planned and unplanned failovers.