Bulk, Cold, vMotion, Replication Assisted vMotion and OS Assisted Migration compared, and the VM attributes that block each one.
From Ultra Transcenders 2V0-15.25 by Tony Rough
Each migration type has its own protocol, downtime and prerequisites. Choosing a type the VM cannot satisfy is the most common reason a migration fails validation. All types need the relevant service active and healthy in the Service Mesh, and for vSphere sources the VM must be in a Service Cluster.
| Type | Technology | Downtime | Parallelism | Key prerequisites |
|---|---|---|---|---|
| Bulk | vSphere Replication, then switchover | About a reboot | Parallel, schedulable | HW version 7 or later. VMware Tools installed. Network Extension for low downtime |
| HCX vMotion | vMotion through the IX | None | Serial between a site pair is the supported model | Tunnels up. HW version 9 or later. x86 |
| Cold | NFC on the vMotion path. Selected automatically for powered-off VMs | VM is off | Must meet the vMotion requirements. IP and MAC kept | |
| Replication Assisted vMotion (RAV) | Replication, then delta vMotion at switchover | None | Parallel replication, serial switchover | Tunnels up. HW version 9 or later. x86. No vVOL. RAV service healthy |
| HCX Assisted vMotion (Direct) | Native cross-vCenter vMotion between ESX hosts | None | Up to 50 concurrent relocations per HCX appliance | Dedicated vMotion path between hosts. Same VDS version. vSphere 6.7 or later. Cross-vCenter vMotion dependencies met |
| OS Assisted Migration (OSAM) | Sentinel agent replication from KVM or Hyper-V guests | Quiesce plus fix-up, two reboots | Up to 50 active replica disks per SGW/SDR or SRG | Sentinel installed in each guest. Guest Network in the Compute Profile |
| Supervisor onboarding | Bulk to a staging area, then import into a Supervisor namespace | Reboot on import | Bulk only | HCX 9.0 or later at both ends. HW version 17 or later. VMware Tools. Guest OS must support guest customisation. No IPv6-dependent workloads |
| Attribute | Bulk | vMotion / RAV | Fix or workaround |
|---|---|---|---|
| RDM physical compatibility | Blocked | Blocked | Not migratable |
| RDM virtual compatibility | Allowed (converted to VMDK) | Blocked | Use Bulk |
| Mounted ISO | Use Force Unmount ISO | Blocked unless Force Unmount ISO is selected | Select the option or detach the ISO |
| Snapshots | Always removed | Fail validation unless Remove Snapshots is selected | Select Remove Snapshots |
| Shared SCSI bus, multi-writer, FT, shared VMDK | Blocked | Blocked | Turn off FT temporarily for RAV |
| DirectPath I/O | Remove the device | Remove the device | |
| vNVMe controller | Blocked | Blocked for RAV | |
| Serial port | Allowed | Blocked | Use Bulk, or detach the device |
| Independent disks | Blocked for RAV (no delta disks) | ||
| VBS | Not supported | Not supported | |
| vSphere VM Encryption | Supported with matching KMS or NKP | Not supported | Bulk with the same KMS name and IP at the target |
| VMware appliances (vCenter, NSX Manager) | Not supported | Not supported | |
| VM that cannot shut down gracefully | Fails unless Force Power-off VM is selected | ||
| VLAN trunk port groups | Not shown in the migration UI | Not shown |
vpxd.certmgmt.mode is set to custom. Deleting the host disrupts those migrations.max_retry_interval in /opt/vmware/hcx/osam/etc/sync.params, 1 to 8 hours, then restart vmware-hcx-osam-sentinel). Linux guests are limited to 15 disks and Windows to 64 volumes.| Option | When | Effect |
|---|---|---|
| Retry | Mobility Group is Failed or Canceled | Creates a new Mobility Group. Seed Checkpoint is on by default for Bulk and RAV, so already-copied data is reused |
| Force Cleanup | Rollback or cancel did not complete. Do this before Retry | Clears processes left at both sites. Can also remove checkpoint data |
| Archive | No further action needed | Removes the migration from the Migration Management and Tracking views and updates the dashboard counters. Log entries are kept |
Seed Checkpoint needs a specific datastore, not a datastore cluster. vCenter alarm codes record outcomes: 65005 (RAV), 65006 (Bulk), 65007 (vMotion) and 65008 (Cold) for errors, and 66001 to 66004 for cancellations. The error alarms direct you to the HCX Migration Tracker.
Common trap: Choosing HCX vMotion to move dozens of VMs with zero downtime - HCX vMotion is supported serially between a site pair. For large-scale zero-downtime moves, RAV replicates in parallel and switches over serially.
Common trap: Retrying a failed migration straight away - if the rollback did not finish, run Force Cleanup first, or the retry and any new migrations can fail on leftover processes.
This note is one section of Ultra Transcenders 2V0-15.25: VMware Cloud Foundation 9.0 Support, an independent study guide that explains every topic the exam covers by technology, with comparison tables, diagrams and the common traps, plus a glossary linked to Broadcom TechDocs.
Free to read with Kindle Unlimited. Opens Amazon in a new tab.
About the book · 2V0-15.25 terms in the glossary · All 2V0-15.25 study notes
Which log on which appliance records the failed step, from domainmanager.log and lcm-debug.log to hostd, vpxa and the NSX logs.
The four starting points, the path and tool each one leads to, and which existing components converge into VCF 9.0.
The two replacement paths side by side, the import rules for external certificates and the documented replacement failures.
What each state means, how hostd, vpxa and FDM fit in, and the first action for a host that is not responding.
What Agent Unreachable, Uninitialized, Host Failed, Network Partitioned and Network Isolated mean, and the documented fix for each.
Absent and degraded compared, object accessibility rules and how vSAN behaves when a device, controller or host fails.
The channels between NSX Manager and a host, and the symptom, cause and fix for preparation and connection faults.