A team must back up and replicate a multi-VM application on AHV so the related VMs can be snapshotted and failed over together as one recoverable set. Which construct should the administrator create?

A protection domain is the Prism Element construct that defines a set of VMs as one snapshot and replication unit. When related VMs are placed in the same protection domain, its schedule takes local snapshots of all of them together and replicates them to the remote site, giving administrators one object to protect, monitor and fail over. Adding the VMs to a consistency group inside the protection domain makes their snapshots crash-consistent with each other. A category on its own is only a label for search and policy targeting; it protects nothing until a protection policy is associated with it, so tagging VMs without a policy creates no snapshots or replication. A storage container or storage policy controls how VM disks are stored, such as replication factor, compression, encryption and QoS, but it doesn't group VMs for snapshots or DR operations. A recovery plan orchestrates power-on order, network mapping and scripts during failover; it works with VMs already protected by a protection policy and does not itself snapshot or replicate anything. Exam caveat: in Prism Central the equivalent pattern is a protection policy associated with a category, plus a recovery plan for orchestration; in Prism Element it is a protection domain. Operational check: add the application VMs to one protection domain, confirm the schedule and remote site, and verify the protection domain reports the expected VM count after the first snapshot.