Combines a witness, cross-cluster live migration and synchronous replication with zero RPO to give AHV both high availability and disaster recovery. If no witness is present, replication stays synchronous but failures are handled by hand or after a timeout.
Read more: Nutanix docs
In the Ultra Transcenders books
Each book explains AHV Metro Availability in context, with comparison tables and the common traps.
Terms in this definition
- Witness
Nutanix's preferred way of handling metro failures. Decisions are automatic, split brain is avoided, and a lost site is distinguished from a broken network link.
- Cross-cluster live migration
Moves running, synchronously protected AHV VMs from one cluster to another as a planned operation with no downtime. The on-demand variant, OD-CCLM, works without a protection policy.
- RPO
Recovery point objective: how much data, expressed as time, you can afford to lose. In a VM backup policy the smallest achievable RPO is the backup frequency.
- Nutanix AHV
The hypervisor from Nutanix. Veeam's integration backs up its VMs to repositories and can restore to it from backups of vSphere, Hyper-V, AHV and other platforms.
- HA
High availability means a service keeps running despite local failures, for instance by spreading replicas across zones.
- DR
Disaster recovery means getting a service running again from another region or site, with success judged by RPO and RTO; uptime SLAs and data retention are different concerns.
Related terms
- Metro availability
Keeps paired storage containers in step through synchronous replication, giving an RPO of zero. Prism Element sets it up on ESXi and Hyper-V 2016, whereas AHV Metro Availability has its own setup in Prism Central.
- Witness Service VM
Supports AHV Metro Availability from a VM independent of Prism Central. Deployments with two Prism Central instances call for this.
- Witness VM
Two-node clusters and ESXi metro availability built on protection domains get their witness from this VM. For AHV Metro Availability, a Witness Service takes the role instead, hosted in Prism Central or on its own VM.