-
Key Takeaways
-
What is V2V Migration?
-
Why Organizations Do V2V
-
How V2V Works: Five Stages
-
Disk Formats Involved
-
The Four V2V Method Families
-
Downtime Expectations by Method
-
Decision Matrix: Which Method Fits Which Scenario
-
Where Backup-Based Migration Fits
-
FAQ
-
Conclusion
V2V migration moves a virtual machine from one hypervisor or virtualization platform to another, for example from VMware ESXi to Hyper-V or Proxmox VE. It works by copying the VM's virtual disks, converting disk format and virtual hardware to what the target platform expects, and adapting the guest operating system (drivers, boot firmware, network settings) so it starts correctly. Methods differ mainly in how much of that conversion they automate and how much downtime the cutover needs.
Key Takeaways
V2V crosses a platform boundary. Moving a VM between hosts of the same platform is not V2V, because no format or driver conversion is needed.
Disk conversion is the easy part. Whether the guest boots depends on firmware mode, storage drivers, and network configuration inside the guest.
There are four method families: OVF/OVA export and import, direct disk conversion, platform-native import tools, and backup-based restore across platforms.
"Live" migration does not mean the same thing across tools. Downtime and rollback behavior depend on how the cutover is designed.
Most V2V methods need at least a short outage. The VM stops at cutover; the length of the outage depends on whether the tool pre-copies data, not on how "live" the method sounds.
Keep the source VM intact until the new VM passes boot, application, and network checks.
What is V2V Migration?
V2V (virtual-to-virtual) migration is the process of importing a virtual machine from one virtualization platform into another. Red Hat's V2V documentation defines V2V as follows and notes that its `virt-v2v` tool also enables paravirtualized VirtIO drivers in the converted guest where possible. (That guide is for RHEL 6 and is legacy; treat it as a definition source, not a current compatibility reference.)
| Term | Source → Target | Conversion needed? |
| V2V | Hypervisor A VM → Hypervisor B VM | Yes: disk format, virtual hardware, guest drivers |
| P2V | Physical server → VM | Yes: hardware abstraction layer changes |
| Live migration (same platform) | Host 1 → Host 2, same hypervisor | No |
| OVF/OVA export | VM → portable package | Packaging only; importer still adapts the VM |
Why Organizations Do V2V
Microsoft's documentation for its VMware-to-Hyper-V conversion extension lists datacenter modernization, test and development validation, and disaster recovery planning as typical scenarios. Other common drivers are consolidating multiple hypervisors, changing licensing or support models, and moving workloads to a platform that fits a new architecture. The motivation matters because it sets the constraints: a one-time exit from a platform tolerates a different level of risk than a recurring test-environment refresh.
How V2V Works: Five Stages
Every V2V method performs the same five stages, whether the user sees them or not.
1. Inventory and pre-check. Record the VM's firmware mode (BIOS or UEFI), disk sizes, number of disks, NICs, static IPs, and installed guest tools.
2. Copy the disks. Data is read from the source through an export, an API connection, or a backup restore point.
3. Convert format and virtual hardware. The disk container changes (for example, VMDK to VHDX or qcow2), and the VM definition is mapped to the target's CPU, memory, controller, and NIC models.
4. Adapt the guest. Storage and network drivers for the new virtual hardware are made available, hypervisor-specific tools are replaced, and boot configuration is aligned.
5. Cut over and validate. The source VM is stopped (or a final delta is synchronized), the target VM starts, and boot, application, and network behavior are checked.
Disk Formats Involved
The `qemu-img convert` command handles conversion among raw, qcow2, VDI, VMDK, and VHD according to the OpenStack image guide. Converting the container does not change what is inside the guest. The same guide warns that a Windows image converted without the VirtIO driver can crash at first boot.
| Platform | Typical disk format | Notes |
| VMware vSphere/ESXi | VMDK | Export as OVF files; a single OVA needs the OVF Tool in vSphere 6.5 and later |
| Microsoft Hyper-V | VHDX (VHD legacy) | Generation 2 boot volumes up to 64 TB; Generation 1 up to 2 TB for VHDX (Microsoft) |
| KVM / Proxmox VE / oVirt / OpenStack | qcow2 or raw | VirtIO drivers are the usual paravirtual path |
OVF itself is a DMTF standard published by ISO as ISO 17203. It standardizes packaging and metadata, but an importer must still translate the VM into its own hardware model.
The Four V2V Method Families
| Method | How it works | Strength | Limitation |
| OVF/OVA export and import | Export a package from the source, import it on the target | Platform-neutral packaging; no direct connection between platforms | Source usually powered off; disk and guest adaptation still needed; manual per VM |
| Direct disk conversion (`qemu-img`) | Convert the disk file, build a new VM shell, attach the disk | Full control; works when no tool connects | Everything else (hardware, drivers, firmware) is manual |
| Platform-native import tool | Target platform reads from source platform and maps config | Automates mapping and some driver handling | Supports specific source/target pairs and versions |
| Backup-based cross-platform restore | Restore a backup of the VM onto a different platform | Reuses existing restore points; source VM is not touched at migration time | Requires a backup product with cross-platform recovery |
Examples of platform-native tools documented by their vendors:
Windows Admin Center VM Conversion extension. Migrates VMs from vCenter to Hyper-V. It makes an initial copy while the source runs, then performs a final delta sync at cutover with the source powered off. Microsoft labels it a preview in that documentation, so verify current status before planning around it. It replaced Microsoft Virtual Machine Converter, whose retirement Microsoft announced in 2016.
Proxmox VE import wizard. Imports ESXi VMs through the Proxmox UI. At launch, Proxmox reported testing against ESXi 6.5 through 8.0; check the current Proxmox documentation for newer versions. Its Migrate to Proxmox VE wiki covers the other import paths.
virt-v2v. Part of libguestfs; converts guests from VMware (vCenter, OVA, VMX) to KVM-based targets such as local libvirt, oVirt, and OpenStack. In its default copying mode, the source guest is left unchanged.
Downtime Expectations by Method
Most V2V migrations need some downtime, because the source VM must stop at some point so the target can start with a consistent copy of the data. What differs by method is when the VM stops and how long it stays down. Exact durations depend on disk size, data change rate, network throughput, and the guest OS, so no method has a universal figure. Rehearse on a non-critical VM to get a real number for your environment.
| Method | Where downtime occurs | Downtime expectation |
| OVF/OVA export and import | From shutdown before export until the imported VM boots on the target | Longest of the four. The VM is normally powered off for the whole export, transfer, and import. Broadcom's export procedure starts with powering off the VM or working from a copy. |
| Direct disk conversion (`qemu-img`) | From the point the disk is quiesced until the converted VM boots | Usually requires a powered-off VM, or a consistent snapshot or clone to convert from. Converting a disk that is still in use risks an inconsistent image. Conversion time scales with disk size. |
| Platform-native import | Cutover window only, if the tool pre-copies; the whole copy, if it does not | Depends on pre-copy and final-sync capability. Microsoft's extension copies while the source runs, then powers the source off for a final delta sync, so downtime is the final sync plus import and boot. Proxmox's live-import requires stopping the source VM first, then starts it on the target while data is fetched on demand, so downtime is short, but the VM depends on the network during the transfer. `virt-v2v` in its default copying mode leaves the source unchanged; downtime depends on whether you stop the source before conversion and for how long. |
| Backup-based restore | Between stopping (or last backing up) the source and starting the restored VM | Depends on restore point age and cutover design. Restoring a point taken before the source stops means changes made since then are not on the target. A final backup after the source is stopped avoids that, at the cost of a longer window. Vinchin documents both full restore for batch migration and instant restore as V2V paths. Whether a VM can start before all data is restored depends on the instant-restore capability and version. |
How to read the table:
Downtime vs. data-loss window. A short outage can still lose data if the copy was taken earlier than the stop. Decide which matters more for each workload.
The phase to plan around. In every method, the unavoidable outage is the final stop, final data sync or restore, and first boot. Boot-time failures (firmware, drivers, network) extend downtime more often than the copy itself, as the first Original Insight explains.
Do not promise a number to stakeholders before a rehearsal. Quote a range from a test run on a representative VM, and keep the source intact so you can roll back if the target fails validation.
Decision Matrix: Which Method Fits Which Scenario
| Scenario | Suggested method | Why |
| A few VMs, downtime acceptable, no direct connection between platforms | OVF/OVA export or direct disk conversion | Simple and portable; manual steps are manageable at small scale |
| vCenter source, Hyper-V target, shorter downtime wanted | Windows Admin Center VM Conversion extension | Pre-copy plus delta; check preview status and guest OS support |
| ESXi source, Proxmox VE target | Proxmox import wizard | Native integration; confirm ESXi and PVE versions |
| VMware, Xen, or OVA source; libvirt, oVirt, or OpenStack target | `virt-v2v` | Built-in guest conversion and VirtIO handling |
| Many VMs across several platforms, or rollback points required | Backup-based cross-platform restore | Migration reuses the restore point; the source is not modified |
Where Backup-Based Migration Fits
Because a backup already contains a copy of the VM, restoring it onto a different platform is a V2V path that does not require reading from the production source again. Vinchin Backup & Recovery documents this as cross-platform recovery for V2V migration, covering both full VM recovery for batch migration and instant VM recovery. Supported platform lists and licensing for this feature vary by version, so confirm the current list on the Vinchin V2V page before planning.
FAQ
Q1: Is OVF/OVA export the same thing as V2V?
No. OVF is a packaging standard, not a conversion process. Broadcom documents that vSphere 6.5 and later export multiple OVF files and need the OVF Tool to build a single OVA. The target still has to convert disks and hardware.
Q2: Can I migrate a running VM?
It depends on the tool and version. Microsoft's extension copies while the source runs but powers the source off for the final sync. Proxmox's live-import requires the source VM to be stopped first, per the Proxmox forum announcement and Thomas-Krenn's guide.
Q3: Do I have to remove VMware Tools?
It depends. Microsoft's extension removes them from Windows VMs after migration; for other tools and Linux guests, check the tool's documentation and verify manually.
Q4: What does V2V cost?
Several tools carry no license fee: `qemu-img` and `virt-v2v` are open-source, and Microsoft describes its extension as available at no extra cost. The real cost is labor, testing, and downtime. Application licensing after hardware or platform change depends on each vendor; check before migrating.
Q5: How do I treat database or stateful VMs?
Use a design with a rollback point (see the second insight above) and plan a controlled application shutdown before the final copy. Tool behavior differs, so rehearse on a non-production copy.
Q6: Is V2V useful for disaster recovery?
Yes. Microsoft lists building Hyper-V replicas of VMware workloads as a DR use case, and backup-based cross-platform restore can bring a VM up on a different platform when the original platform is unavailable. Test this path before you need it.
Q7: How do I know the migrated VM is trustworthy?
Verify package checksums where the format provides them (OVA manifests), compare application test results before and after, and review security configuration against your baseline, as NIST SP 800-125 recommends for virtualized environments.
Conclusion
V2V migration succeeds or fails inside the guest, not in the disk file. Pick a method by what it adapts automatically and how its cutover behaves when something goes wrong, then rehearse on a non-critical VM first. Keep the source until boot, application, and network checks pass. Backup-based restore is one option when rollback points also matter.
Share on: