Toil-Less

Guide · VMware → Hyper-V

VMware to Hyper-V: what to check before you migrate

For many Windows shops, Hyper-V is the obvious way off vSphere. The licences are often already paid for, the team knows Windows, and Failover Clustering is mature. Most VMs convert without drama. The ones that don't are usually the same few kinds, and you can find them before you move anything. This checklist covers what blocks a VM, what needs preparation, and what you will have to rebuild around the VMs.


1. Start with an inventory

Before you plan migration windows, know what you are moving:

  • How many VMs, and which of them are VMware's own infrastructure (vCenter, vCLS, other appliances). Those are retired with vSphere, not migrated.
  • Used storage, not provisioned storage. Copy time depends on the data that actually exists.
  • Guest operating systems and their support status. A VM on an operating system that is already out of vendor support is a candidate for replacement, not migration.
  • What you will need on the target: total vCPUs, RAM and storage, plus headroom for a failed node if you build a cluster.
  • Licensing. Windows Server Datacenter covers unlimited Windows Server VMs on a licensed host. Standard covers two per fully licensed host. Count what you actually need before you size the cluster.

2. Blockers: VMs that need their own plan

Blocker Why it matters
Physical-mode RDMs There is no virtual disk to convert, and image-based backups don't include it. Re-present the storage to the guest (for example via in-guest iSCSI) or copy the data at file level into a VHDX.
Shared multi-writer disks Typical of guest clusters. Hyper-V uses shared VHDX (VHD Set, .vhds) for this, so the shared storage has to be rebuilt.
PCI or GPU passthrough Hyper-V offers Discrete Device Assignment (DDA), and recent versions add GPU partitioning, but both depend on the host hardware. Plan the target host first.
vSphere VM Encryption Encrypted VMs must be decrypted in vSphere before any conversion tool can read their disks.
Fault Tolerance There is no lock-step equivalent. Failover Clustering restarts a VM on another node after a failure, and Hyper-V Replica is asynchronous replication, not zero-downtime protection. Decide whether a short outage is acceptable.

3. Preparation: fix these before migration night

  • Snapshots. Consolidate or delete them first. They break or slow down most conversion paths and ruin copy-time estimates.
  • Independent disks. They are skipped by snapshots, and therefore by snapshot-based tools and backups. Check where their data should go.
  • USB devices, typically licence dongles. Hyper-V has no USB passthrough for VMs. Plan a network USB server before you move these VMs, not after.
  • Serial ports. Hyper-V maps a VM's COM ports to named pipes, not to physical ports. Check what is attached and whether it still matters.
  • vTPM. Generation 2 VMs support a virtual TPM, but its contents don't migrate. If the guest uses BitLocker, have the recovery keys ready or suspend BitLocker first.
  • Mounted ISO images. Disconnect them.
  • Memory limits below the configured RAM. Usually an old misconfiguration. Decide on Dynamic Memory or static RAM on the target instead.
  • VMware Tools not installed or not running. Without Tools you have no reliable hostname, IP or exact OS version for that VM, so check it by hand.

4. Firmware decides the VM generation

This is the Hyper-V detail that catches most people:

  • BIOS VMs become Generation 1 VMs. The boot disk sits on an IDE controller, which limits it to 2 TB.
  • UEFI VMs become Generation 2 VMs, with SCSI boot, Secure Boot and vTPM support.
  • You cannot change a VM's generation after it is created. Converting a BIOS guest to UEFI is an in-guest job (for Windows, mbr2gpt) that should happen before or instead of the move, not during it.
  • Linux on Generation 2 with Secure Boot needs the Microsoft UEFI Certificate Authority template, not the default Windows one.

5. Windows guests

  1. Uninstall VMware Tools before migrating. Removing it afterwards, on a host that is no longer VMware, is often difficult.
  2. Document static IP settings. The new synthetic network adapter is a new adapter for Windows. The old configuration stays on the old, now hidden adapter.
  3. Integration Services are built into current Windows versions and updated through Windows Update. Very old guests may need attention.
  4. Expect re-activation for some Windows and application licences tied to hardware IDs.

6. Linux guests

  • Current distributions ship the Hyper-V drivers (hv_vmbus, hv_storvsc, hv_netvsc) in the kernel. Older or minimal installations may not have them in the initramfs, so check before the move.
  • Network interface names change. Static configurations then point at an interface that no longer exists. Plan to fix this on first boot.
  • Replace open-vm-tools with the Hyper-V daemons package of your distribution.

7. Networks you have to rebuild

Hyper-V handles VLANs differently from vSphere:

  • On vSphere, the port group carries the VLAN. On Hyper-V, there are no port groups: a VM's network adapter carries the VLAN ID (Set-VMNetworkAdapterVlan -Access -VlanId) on an external virtual switch.
  • So list every port group that is in use, its VLAN ID, and which VMs depend on it. That list becomes the VLAN setting on each VM's adapter.
  • Look out for port groups with different VLAN IDs on different hosts. On vSphere it means a VM's network depends on which host it runs on. Clear it up before you rebuild it.
  • For teamed uplinks on current Windows Server, Switch Embedded Teaming (SET) is the usual choice.

8. Placement rules you have to rebuild

DRS rules don't come along. Write them down before you shut vCenter off:

  • Keep apart rules (for example two domain controllers) map to the cluster group property AntiAffinityClassNames. It is a soft rule by default. Setting the cluster's ClusterEnforcedAntiAffinity to 1 makes it a hard rule.
  • VM-to-host rules: a should run on these hosts rule (soft) maps to preferred owners on the cluster role, a must run on these hosts rule (hard) to possible owners.
  • Keep together rules: Windows Server 2022 and later have cluster affinity rules in PowerShell (New-ClusterAffinityRule) with a SameNode type, which keeps VMs on the same node, and a DifferentNode type for keeping them apart. They are managed in PowerShell, not in Failover Cluster Manager. On Windows Server 2019 and earlier there is no direct equivalent.
  • Failover Clustering can balance VMs across nodes (VM Load Balancing), but it is simpler than DRS. Check its settings rather than assuming DRS behaviour.

9. Backups

If you use Veeam:

  • Independent disks are not included in Veeam's image-based VMware backups. If you plan a backup-and-restore migration, handle those disks separately.
  • VMware Tools status matters for application-aware processing. Outdated or missing Tools can make those jobs fail.
  • Veeam supports Hyper-V natively, and restoring VMware backups to Hyper-V is a common migration path. It also gives you a fallback copy of every VM.

10. Estimate copy time realistically

Rule of thumb: used storage divided by about 70% of your link speed.

Used storage 1 Gbit/s 10 Gbit/s
500 GB about 1.6 hours about 10 minutes
2 TB about 6.3 hours about 38 minutes
10 TB about 32 hours about 3.2 hours

This is pure transfer time. Conversion, testing and cut-over come on top. Disks larger than 2 TB deserve their own window.

11. Choose the migration path

  • System Center Virtual Machine Manager, if you already run it, can convert VMware VMs. Microsoft currently points customers to it for VMware to Hyper-V migrations.
  • VM Conversion in Windows Admin Center Virtualization Mode (vMode). Microsoft's newer tool. It places a Windows appliance VM on each source ESXi host, syncs disks with changed block tracking while the VM keeps running, and migrates up to 10 VMs per batch. As of November 2026 it is in private preview, available only through Microsoft's Windows Server Technology Adoption Program (TAP). The earlier VM Conversion extension for Windows Admin Center has been removed from the extension feed, so guides that point to it are out of date. Check the current status before you plan with it.
  • Backup and restore with Veeam or another tool that supports both platforms.
  • Third-party converters for VMDK to VHDX, for edge cases.

Whatever you use, migrate a few unimportant VMs first and test them properly before you move anything that matters.


The short version

Before migrating, know for every VM:

  • ☐ Does it have a blocker: physical RDM, shared disk, passthrough, encryption or Fault Tolerance?
  • ☐ Are snapshots consolidated and ISOs disconnected?
  • ☐ Does it use USB devices or serial ports?
  • ☐ BIOS or UEFI, so Generation 1 or 2? Secure Boot or vTPM in use?
  • ☐ Windows: IP settings documented, VMware Tools removed?
  • ☐ Linux: Hyper-V drivers present, network interface names handled?
  • ☐ Which VLANs does it need, and which placement rules apply?
  • ☐ Is the operating system still supported?
  • ☐ How much used storage does it have, and how long will it take to copy?

Check all of this automatically

The VMware Exit Readiness Report reads your vCenter and runs these checks for every VM: blockers, preparation items, OS support, networks, DRS rules and copy time. It produces one HTML report with a summary for your manager. Read-only, plain PowerShell, one-time price.

See the report → View an example report