HCX Replication Assisted vMotion (RAV) uses the HCX along with replication and vMotion technologies to provide large scale, parallel migrations with zero downtime.
HCX RAV provides the following benefits:
Large-scale live mobility: Administrators can submit large sets of VMs for a live migration.
Switchover window: With RAV, administrators can specify a switchover window.
Continuous replication: Once a set of VMs is selected for migration, RAV does the initial syncing, and continues to replicate the delta changes until the switchover window is reached.
Concurrency: With RAV, multiple VMs are replicated simultaneously. When the replication phase reaches the switchover window, a delta vMotion cycle is initiated to do a quick, live switchover. Live switchover happens serially.
Resiliency: RAV migrations are resilient to latency and varied network and service conditions during the initial sync and continuous replication sync.
Switchover larger sets of VMs with a smaller maintenance window: Large chunks of data synchronization by way of replication allow for smaller delta vMotion cycles, paving way for large numbers of VMs switching over in a maintenance window.
HCX RAV migration triggers the following events:
Replication begins with a full synchronization (replication) of the virtual machine disks to the destination site.
Migrated VMs enter a continuous synchronization cycle until a switchover is triggered.
- Depending upon data churn on source disk, additional snapshots are being created during the RPO cycle. After each RPO cycle, disk consolidation takes place and creates a “hbrdisk.RDID vmdk” called as replica instance vmdk file on target datastore. Refer VMware KB 87028.
You can have the switchover process start immediately following the initial sync or delay the switchover until a specific time using the scheduled migration option. If the switchover is scheduled, the synchronization cycle continues until the switchover begins.
The final delta synchronization begins when the switchover phase starts. During this phase, vMotion is engaged for migrating the disk delta data and virtual machine state.
As the final step in the switchover, the source VM is removed, and the migrated VM is connected to the network powered on.
Replication Assisted vMotion creates two folders at the destination site. One folder contains the virtual machine infrastructure definition, and the other contains the virtual machine disk information. This is normal behavior for RAV migrations and has no impact on the functionality of the virtual machine at the destination site.
In some cases, having two folders might impact other applications, such as backup tools, that require access to the virtual machines folders. If necessary, you can consolidate the contents of these two folders using VMware Storage vMotion.
Requirements for Replication Assisted vMotion
- HCX Interconnect tunnels must be up/active.
- HCX vMotion requires 100 Mbps or higher throughput capability.
- The virtual machine hardware version must be Version 9 or higher.
- The underlying architecture, regardless of OS, must be x86.
- Hybrid Interconnect, Bulk Migration, vMotion, and Replication Assisted vMotion services must be activated and in a healthy state in the relevant Service Mesh.
- The resources to create, power on and use the virtual machine must be available in the destination environment.
- Virtual machines must reside in a Service Cluster (defined in the Compute Profile).
- RAV uses vSphere Replication whose potential throughput can vary depending on bandwidth available for migrations, latency, available CPU/MEM/IOPS, and disk read speed. For more information about how to determine bandwidth requirements, see Bandwidth Requirements for vSphere Replication.
- VMware NFC is used as a secondary protocol during HCX Replication Assisted vMotion migration.
HCX always uses its Management interface NFC traffic. In deployments where ESXi servers use a dedicated Provisioning vmkernel for NFC traffic, HCX continues to route NFC traffic through the Management interface.
For information regarding the migration limits, search for VMware HCX in the VMware Configurations Maximums tool.
Restrictions for Replication Assisted vMotion
Virtual machines with the following attributes are not supported for migration.
Virtual machines cannot be migrated while they are using a shared SCSI bus, flagged for multi-writer, enabled for fault tolerance, or configured for shared VMDK disk sharing.
Attached virtual media or ISOs.
Virtual Machine Hardware Version 8 or earlier.
Live switchover of concurrent RAV migrations is run serially.
HCX vMotion defaults to Opportunistic mode for per-VM vMotion Encryption if it is set to Required. During the migration operation - the mode is changed to Opportunistic on the migration initialization, and then set back to Required after the migration is completed.
HCX Replication Assisted vMotion does not support the migration of workloads with Independent persistent and Independent non-persistent disks, as taking a snapshot of such virtual machine does not produce delta disks, which are required for the underlying RAV migration technology.
Virtual machines with Raw Device Mappings (RDM) in Physical Compatibility mode cannot be migrated.
- Virtual machines with Raw Device Mappings in compatibility mode (RDM-V) cannot be migrated using RAV and vMotion.
Virtual machines with DirectPath I/O configurations cannot be migrated without first removing the DirectPath device.
To migrate FT-enabled VMs, temporarily turn off Fault Tolerance, and perform RAV. When this operation is complete, turn Fault Tolerance back on.
- Migration to or from vVOL datastores is not supported.
- The HCX Migration interface does not display port groups that are VLAN trunks.
Virtual machines that cannot be gracefully powered off cannot be migrated. HCX can override with the Force Power-off VM option.
Virtual Machines with Change Block Tracking (CBT) can be migrated, but HCX deactivates CBT.
RAV migration to VMFS6 target datastores requires the following minimum vSphere version at the target site: vSphere 6.5U3f or vSphere 6.7U3.
Virtual machines using virtual NVMe (vNVME) Controllers cannot be migrated.
Virtual machine Snapshots cannot be migrated.
If either the source or the destination vCenter version is earlier than 6.0, VMware vCenter tags attached to a virtual machine are not retained during migration. Tags can be created manually for those virtual machines after migration.
VMware software appliances (including, but not limited to vCenter Server and NSX Manager) cannot be migrated with HCX RAV migration.
- vSphere VM Encryption is not supported with HCX migrations.
- Virtual machine workloads with an attached serial port device are not supported. For these workloads, detach such devices prior to starting the migration.
Note: You can use Bulk Migration to migrate workloads with an attached serial port device.
- Virtual Based Security (VBS) functionality is not supported with VMware HCX migrations.