Data Protection and Resiliency
This domain covers protecting VMs locally and remotely: snapshots, protection policies, replication and orchestrated recovery.
Local protection
- Snapshots are crash-consistent by default; installing Nutanix Guest Tools enables VSS/application-consistent snapshots and self-service restore.
- Protection policies in Prism Central define schedule, retention and optionally a remote target; consistency groups keep multi-VM applications in a single snapshot.
- Snapshots on the same cluster protect against VM-level mistakes, not against cluster or site loss.
Remote protection
- Asynchronous replication ships snapshots to a remote cluster; your RPO equals the schedule, not the available bandwidth.
- NearSync uses continuous replication with an RPO measured in minutes (as low as about one minute) and needs low-latency links.
- Metro Availability is synchronous across a stretched cluster with RPO zero; a witness at a third site breaks the tie during a site failure.
- Nutanix Leap executes recovery plans that boot VMs in dependency order and supports test failover into an isolated network.
Worked example
Fifty VMs, RPO 15 minutes, RTO one hour. Choose a protection policy with a 15-minute snapshot schedule replicated to the DR cluster, install NGT for application consistency, and build a Leap recovery plan with a test run. Reasoning: a 15-minute RPO rules out nightly replication; a one-hour RTO is achievable by booting replicated snapshots in dependency order; the test proves the plan without touching production.
Traps
- Treating snapshots as backups - a cluster loss takes the local snapshots with it.
- Promising an RPO the schedule cannot deliver, or ignoring bandwidth and latency for NearSync.
- Placing the witness inside one of the two protected sites, which defeats its purpose.
- Building a recovery plan but never running it in test mode.