Live-Migrating VMs Between Nutanix AHV Clusters

Diagram of a running VM live-migrating between two Nutanix AHV clusters managed by Prism Central

On-Demand Cross-Cluster Live Migration (OD-CCLM) lets you move a running AHV VM, or a batch of them, to a different cluster straight from Prism Central. You don’t need a protection policy or any disaster recovery setup: pick a VM, pick a destination, and migrate.

The feature first shipped in AOS 6.7, and each release since has improved it. This guide covers the current release, Prism Central 7.6, AOS 7.6, and AHV 11.2 (July 2026), including what has changed in the 7.x line.

How it works

OD-CCLM uses the same synchronous-replication technology as Nutanix Disaster Recovery, but it sets that replication up for you. A migration runs in three phases:

Migration checks. Prism Central confirms the destination has enough CPU, memory, storage, and (where relevant) GPU capacity, and flags anything unusual before you commit.
Storage synchronization. The VM’s disks are copied to the destination in the background, and every new write is mirrored to both clusters synchronously so nothing is left behind.
VM migration. The remaining state (memory, CPU and device state) moves across, and the VM keeps running on the destination cluster.

What’s new in the 7.x releases

Older AHV to newer AHV (7.6). You can now live-migrate from a cluster running an older AHV version to one running a newer version with no VM downtime. The move is one-way, from older to newer. This helps when you’re upgrading: build or upgrade the new cluster first, then drain the old one onto it.
Different storage container names (7.5). The source and destination clusters no longer need matching container names.
Flow Network Security (7.5). VMs keep their Flow security policies across the move without traffic disruption.

Requirements

Everything below assumes the current release. For older releases, check the OD-CCLM Requirements and Limitations pages in the Prism Central Guide for your version.

Software versions: Prism Central 7.6, with AOS 7.6 and AHV 11.2 on both clusters, is the recommended baseline. Since 7.6 the destination can run a newer AHV version than the source, but not an older one.
Prism Central scope: Both clusters are registered to the same Prism Central, or to Prism Centrals that are paired as availability zones.
Network latency: Nutanix recommends ≤ 5 ms round-trip latency between clusters and supports a maximum of 40 ms. Every write is mirrored synchronously during the move, so latency directly affects VM performance while the migration runs.
Bandwidth: Plan for enough throughput to copy the VM’s disks and memory in a reasonable window. In practice that means more than one active 10 GbE link, such as bonded 10 GbE or 25 GbE+, especially for large or write-heavy VMs.
CPU compatibility: The destination hosts must support the CPU features the VM is using. The migration checks catch mismatches, and clusters on very different CPU generations are the most common reason a migration gets blocked.
Network mapping: The VM’s networks must map to equivalent networks (VLANs/subnets) on the destination so it keeps its connectivity after the move.
CVM SSH access: OD-CCLM uses SSH between the CVMs. If you’ve hardened your clusters by disabling CVM SSH, check the security guidance for your release before relying on this feature.

Step-by-step

Step 1: Check the environment

Confirm the versions, Prism Central registration, latency and CPU compatibility above. Check that both clusters are healthy (NCC checks pass, no ongoing upgrades or rebuilds).

Step 2: Prepare the VMs

Note each VM’s networks, and confirm matching networks exist on the destination.
Check the VM against the Limitations page for your version, for example specific passthrough devices or special VM configurations.

Step 3: Start the migration

Log in to Prism Central and open the VMs list.
Select one VM or several, then choose Actions → Migrate Across Clusters.

Step 4: Configure the destination

Select the destination availability zone (if you have paired PCs) and the target cluster.
Map each source network to its destination network.
Review the results of the pre-migration checks and fix any warnings.

Step 5: Run and monitor

Click Migrate. Track progress from the VM’s task in Prism Central.
If you want to prove there’s no downtime, keep a continuous ping to the VM running during the move.

Step 6: Post-migration checks

Confirm the VM is running on the destination cluster and that its applications respond.
Check cluster-specific settings such as categories, protection policies and affinity rules, and confirm your Flow security policies were applied as expected.
Update your records, including CMDB, backup jobs and monitoring targets that reference the old cluster.

When to use it

Hardware refresh / cluster retirement: drain an old cluster onto a new one without a maintenance window. With 7.6 the new cluster can already be on a newer AHV version.
Rebalancing: move a hot workload to a cluster that has more headroom.
Planned events: move VMs off a site ahead of power or facilities work.

OD-CCLM is not a DR mechanism. It needs both clusters online and is intended for planned moves. For unplanned outages, you still need a protection policy and a recovery plan.

Conclusion

On-Demand Cross-Cluster Live Migration makes moving VMs between clusters a routine operation instead of a project, and in Prism Central 7.6, AOS 7.6 and AHV 11.2 it works across AHV versions too. The requirements that matter most are current versions, Prism Central scope, latency within 5 ms (40 ms maximum) and compatible CPUs.