What's inside the field guide
"Migrating from VMware with fewer than 10 hosts" is a field guide from Vates (October 2026) for small IT teams:
3 to 10 hosts, a few dozen to a couple hundred VMs, and one or two people running IT without a project team. The
guide covers a VMware-to-Vates VMS migration end to end: a two-day audit, a lab on spare hardware, the VM-by-VM
migration routine, storage and backup decisions, the six most common failure modes, a three-week calendar and a
cost comparison.
Why small VMware estates are moving now
Since the Broadcom acquisition, VMware perpetual licenses are gone and everything is a subscription billed per
core, with a 16-core minimum per CPU: a host with two 10-core processors is billed for 32 cores. Bundles include
components such as vSAN, NSX and Aria that small structures never deploy, and the VVF bundle was withdrawn from
sale in several European countries in late 2025. vSphere 7 has been out of support since October 2025 and
vSphere 8 follows in October 2027. Because vSphere 8 dropped support for older CPU generations and I/O devices,
many small estates also face a hardware refresh, and the guide notes that those new servers can just as well
become the first XCP-ng hosts.
What Vates VMS replaces
Vates VMS is an open source virtualization platform from Vates, the French vendor behind XCP-ng and Xen
Orchestra. XCP-ng replaces ESXi as a type-1 hypervisor built on the Xen Project. Xen Orchestra replaces both
vCenter and the backup product in one web console, with no agents inside the VMs. XO Lite, embedded in every
XCP-ng host, has no VMware equivalent and can serve as the main interface for a 3-host shop.
- The Vates VMS subscription is per host, not per core, socket or VM.
- An expired subscription never stops the VMs: you lose the support safety net, not your infrastructure.
-
Support is staffed by the engineers who build the product, up to 24/7 with a one-hour response commitment on
critical incidents.
-
There is no reduced small-business edition: live migration and high availability are in every deployment.
The two-day audit
The guide treats a two-day audit as the cheapest insurance on the whole migration. Day 1 is inventory: every VM
with its OS, resources, disk sizes, datastore type (NFS, VMFS 5, VMFS 6 or vSAN) and actual owner, plus disks
over 2 TB, long snapshot chains and any software touching the hypervisor (backup agent, monitoring, antivirus
console, UPS shutdown tool). Day 2 is constraints: backup obligations, the network and its VLANs, hardware
compatibility with XCP-ng, the two or three VMs that cannot take an evening of downtime, and the VMware renewal
date. The guide also recommends test-restoring one backup on the current platform before anything moves.
A lab that becomes your first host
The lab runs on any 64-bit server with virtualization extensions. XCP-ng installs from a USB stick in about
fifteen minutes, and nothing costs anything until production. The guide lists three things to validate: import a
real Windows VM through the V2V wizard (the XCP-ng Windows drivers are Microsoft-signed), break something on
purpose, and run a backup to the NAS followed by a full restore. With storage on a NAS or SAN, the lab machine
becomes the first production XCP-ng host and each emptied ESXi host is reinstalled as the next one, so no extra
hardware is needed. A vSAN source or an XOSTOR target breaks this rotation and requires planning the XCP-ng side
as a block.
How the V2V migration works
Xen Orchestra includes a V2V (VMware to Vates) wizard that connects to ESXi or vCenter over the API, with no
agent and no direct storage access. On NFS and most VMFS 5 datastores, migration is warm: an initial sync runs
while the VM keeps serving users, then a short cutover transfers only the changed blocks, so downtime is
minutes. On VMFS 6 (the default since vSphere 6.5) and vSAN, the VM is shut down for the whole copy and downtime
follows disk size.
The per-VM routine is short: consolidate snapshots, uninstall VMware Tools, launch the import, trigger the
cutover at an agreed time, install the XCP-ng guest tools and check the application with its owner. XAPI tasks
time out after 24 hours by default, so the limit must be raised for longer transfers. The source VM stays intact
on ESXi, which makes rollback a matter of restarting the original.
Storage and backup decisions
The guide describes three storage situations at this scale. A NAS (NFS or SMB) is kept and mounted as a new
XCP-ng repository, with free space for both copies during migration. A small SAN (iSCSI or Fibre Channel) gets a
new LUN as repository. An aging array, or no shared storage, is the case for XOSTOR, which pools local disks
into a replicated virtual SAN from three hosts. Virtual disks up to 16 TB are supported on almost all repository
types, with XOSTOR and SMB as current exceptions.
Xen Orchestra provides incremental backups with changed block tracking, replication, full-VM and file-level
restore, targets on NFS, SMB or S3-compatible storage, and health checks that boot-test backups. The
non-negotiable rule is to run the old and new backup chains in parallel and retire the old product only after
its last retention window.
What can go wrong, and a three-week calendar
The guide lists six failure modes with fixes: a VM that will not boot (driver or BIOS/UEFI mismatch), a slow
initial sync, a snapshot chain sabotaging the transfer, an appliance VM that resists, a hidden dependency, and
being stuck alone. For an estate of 3 to 6 hosts and 30 to 80 VMs, the project runs about three weeks, or 8 to
12 working days: 2 days of audit, a few lab sessions, 2 to 3 evenings of pilot VMs, 1 to 2 evenings per batch,
and 1 to 2 days of decommissioning. Critical VMs go mid-project, and one full business cycle passes before any
source is deleted.
What it costs
In the guide's worked example, 5 hosts with 2 CPUs of 16 cores each are billed as 160 cores by VMware and 5
hosts by Vates VMS. At the lowest public reseller list prices (checked 5 August 2026, no negotiated discount),
the three-year total is €92,520 for VMware VVF against €13,500 for Vates VMS Pro, a difference of €79,020. The
guide states that the gap stays above 3x even at half the VMware list price, and that moving to 32-core CPUs
would double the VMware bill while the Vates line stays flat.
Key takeaways
- Two days of audit determine whether the rest of the migration goes smoothly.
- Downtime is minutes per VM on NFS and VMFS 5, and the full copy time on VMFS 6 and vSAN.
- Rollback is available until you delete the source VM on ESXi.
- With a NAS or SAN, the migration needs no extra servers.
- No subscription is due until production, and Vates VMS is billed per host.
Download the field guide (PDF)