SAMSIGsince 1979

, ,

A Practical Look at OpenShift Virtualization Engine for VMware Refugees

Title graphic: OpenShift isn't the beast you think – a practical, honest look at OpenShift Virtualization Engine (OVE)

The short version

If you are moving off VMware and only need to run VMs, Red Hat’s OpenShift Virtualization Engine (OVE) deserves a place on your shortlist next to Proxmox and Hyper-V. It is less of a beast than its reputation, and VM migration tooling comes with it. It is also not as simple as Proxmox, and this article is honest about where.

Since Broadcom bought VMware and moved to subscription-only licensing, a lot of shops are re-platforming. The options I see most often are Hyper-V, because many shops already own Windows Server licenses, and Proxmox, because it looks and feels like ESXi: download an ISO, install, open a web UI, create a VM. Proxmox is a genuinely good choice and I am not going to argue otherwise.

OpenShift, on the other hand, rarely makes the list. Mention it to a seasoned sysadmin and you hear “that’s Kubernetes, that’s for developers, that’s complicated.” Part of that is fair. Part of it is that containers are still an unfamiliar abstraction for many of us who grew up on vCenter. The point of this article is to show that, for plain virtualization, you don’t need to become a Kubernetes expert to get value out of OpenShift.

What OVE actually is

OVE is OpenShift with the container platform features stripped from the subscription, sold for running VMs only. Red Hat describes it as a cost-effective, virtualization-only edition that leaves out container-based and cloud-native application features (Red Hat docs). A Red Hat product manager put it plainly at launch: it is the same OpenShift Virtualization code as the other editions; the difference is the consumption model (The Next Platform).

Under the hood, the hypervisor is plain KVM, managed through the open-source KubeVirt project (Red Hat blog). That matters for anyone worried about lock-in: it is the same KVM that Proxmox uses. Each VM runs inside a pod, but in daily use you mostly see a VM list, a console and a settings page.

Diagram of the OVE stack: your VMs on top, then the web console, KubeVirt, Red Hat CoreOS with KVM, and bare-metal servers; alongside are the Migration Toolkit for Virtualization, storage options (ODF, Portworx, SAN/NAS CSI) and backup vendors (Veeam, Commvault, Cohesity, Rubrik)
The OVE stack in one picture: KVM at the bottom, your VMs at the top, and the ecosystem around it.

What you get in practice:

  • A virtualization-focused web console. Create VMs from templates, open a VNC or serial console, snapshot, clone, live-migrate between nodes.
  • Windows and Linux guests with vendor certification. OpenShift Virtualization is certified in Microsoft’s SVVP program for Windows Server workloads (release notes), and Red Hat lists Windows Server 2016 through 2025 and Windows 10/11 as certified guests (Red Hat KB).
  • The Migration Toolkit for Virtualization (MTV) to pull VMs in from other platforms (more below).
  • A path to containers. If you want them, you switch to a different OpenShift edition rather than migrating again (Red Hat docs).

You are allowed to run infrastructure services in containers on OVE, but not your own application containers. Those would need separately licensed OpenShift clusters (IBM docs).

OVE vs Proxmox, honestly

Proxmox wins on simplicity and price. OVE’s strengths are migration breadth, its partner ecosystem and a future path to containers. Both vendors sell enterprise support, so compare coverage hours, response commitments and procurement fit rather than assuming one side wins. Both run KVM underneath, but that does not make performance equal: benchmark representative workloads, including storage, networking and platform overhead.

Proxmox VEOpenShift Virtualization Engine
HypervisorKVMKVM (via KubeVirt)
Cost to run without supportFree, full feature setPaid; 60-day trial, or one year at no cost via Red Hat Level Up for eligible existing Red Hat customers
Support subscription€550 per socket/year Standard, €1,100 Premium (wz-it); Premium 24/7 since 19 October 2026, Standard 24/7 later in Q4 2026 (StorageReview)From about €1,760 per node/year Standard, node = 1–2 sockets, up to 128 cores (reseller-advertised price)
Smallest installOne node, ISO installOne node (single-node OpenShift), no HA
Smallest HA cluster3 nodes (or 2 + QDevice)3 nodes, or 2 nodes + small arbiter node
Built-in VMware importESXi 6.5–8.0 import wizard, OVF/OVAMTV: vSphere warm/cold, OVA
Other import sourcesManual disk importRHV, OpenStack, other OpenShift clusters; Hyper-V in Technology Preview
Learning curve for a VMware adminLowModerate: some Kubernetes concepts
VendorProxmox Server Solutions (Austria)Red Hat (IBM)

Prices were found in October 2026 and will differ by region and partner. The Proxmox figures are published list prices; the OVE figure is a reseller-advertised price, not an official Red Hat list price, so get a real quote. Note that the OVE price per node covers up to 128 cores, which Red Hat markets as double the previous bare-metal entitlement (Red Hat). On dense modern servers that narrows the gap, but Proxmox remains clearly cheaper.

One licensing detail people miss: OVE does not include RHEL guest entitlements, while the bigger OpenShift editions include unlimited RHEL VMs (Trilio). If you run many RHEL guests, price the full OpenShift Container Platform too.

Getting started: a lab in an afternoon

The install is a web wizard plus a boot ISO, not a pile of YAML. Red Hat’s Assisted Installer does the heavy lifting, and OVE’s own docs name it as the install method, with a bundle that pulls in OpenShift Virtualization and its supporting operators in one tick (Red Hat docs).

  1. Pick your route. For a first look on one box, create a free Red Hat account and build a single-node cluster in the Assisted Installer. It runs as a time-limited, self-supported evaluation without Red Hat support (Red Hat docs), using the same OpenShift Virtualization code that OVE ships. This homelab walkthrough (community blog) shows the whole process. For a proper proof of concept, the official OVE trial runs 60 days on three bare-metal nodes and includes Customer Portal access and Advanced Cluster Management for Virtualization (Red Hat trials). Existing Red Hat customers should also check Level Up, which currently offers a year of fully supported OVE at no cost.
  2. Pick hardware. For a first look, one bare-metal box is enough. Single-node OpenShift needs at least 8 vCPUs, 16 GB RAM and 120 GB disk; with virtualization enabled you also need a second local disk of at least 50 GiB for VMs (Red Hat docs). Treat those as floors; 32 GB+ RAM makes room for actual VMs.
  3. Prepare DNS. You need records for the API and a wildcard for the apps domain. This is the step VMware admins most often trip on, because ESXi never asked for it.
  4. Create the cluster in the Assisted Installer at console.redhat.com: name, base domain, choose single node or multi-node, and select the virtualization bundle.
  5. Boot the discovery ISO on your server(s). The hosts call home, the wizard runs pre-flight checks on CPU, RAM, disks and network, and shows you any potential error before anything is installed.
  6. Click install. The node reboots a few times on its own. When it finishes you get a console URL and an admin password.
  7. Create a VM. In the console, go to Virtualization, pick a template (RHEL, Fedora, Windows), set CPU and memory, start it, and open the console tab. This part will feel familiar.

For production you would install a three-node cluster instead. Since OpenShift 4.20 there is also a supported two-node option, where a small third “arbiter” machine only holds cluster quorum and runs no VMs (Red Hat blog). Plan capacity for it: if one of two nodes fails, half the load has to move, not a third (Red Hat blog).

Migrating VMs: the Migration Toolkit for Virtualization

Migration tooling is included, but it is an operator you install, not something already switched on. MTV is a one-click install from OperatorHub and gives you a wizard: connect source and destination, map networks and storage, build a plan, run it (Red Hat Developer). It is part of what you get with OVE (The Next Platform).

Diagram: VMware vSphere (warm or cold), Red Hat Virtualization, OpenStack and OVA are supported sources and Hyper-V is a Technology Preview; all flow through the Migration Toolkit for Virtualization into OpenShift Virtualization Engine
Supported migration sources in MTV today. Hyper-V is still a Technology Preview.

Supported sources in MTV 2.10 are VMware vSphere, Red Hat Virtualization, OpenStack, OVAs exported from vSphere, and other OpenShift Virtualization clusters (MTV 2.10 docs). MTV 2.12 adds a Hyper-V provider, including Windows failover clusters, but as a Technology Preview that Red Hat does not recommend for production (MTV 2.12 release notes).

Two migration modes matter for VMware users:

  • Cold migration is the default: the source VM is shut down while its disks are copied (Red Hat docs).
  • Warm migration copies disks while the VM keeps running, using VMware changed block tracking, then does a short cutover. CBT must be enabled on the VMs and their disks (Red Hat docs).

The conversion itself uses virt-v2v, which swaps in KVM drivers so guests boot on the new platform (MTV 2.10 planning guide). For comparison, Proxmox’s ESXi import wizard is simpler to start with and supports a live-import mode that cuts downtime, but it only targets ESXi and needs manual driver work for Windows guests (Proxmox forum, computingforgeeks).

Beyond plain VMs: AI tooling, VDI and multi-platform management

Because OVE is OpenShift underneath, it plugs into an ecosystem that a standalone hypervisor doesn’t. A few examples worth knowing, with their current status:

  • AI agents via MCP. Red Hat ships an MCP (Model Context Protocol) server for OpenShift, so AI assistants can work with a cluster under RBAC rules you define. It is a Technology Preview, and its OpenShift Virtualization toolset for VM lifecycle and troubleshooting is still a Developer Preview (Red Hat blog). The upstream docs list tools such as creating VMs and reporting on the virtualization operator’s health in natural language (OKD docs). MTV 2.10 also added MCP support for migration insights as a Developer Preview (Pronteff). Interesting to experiment with, not yet something to run production on.
  • VDI with Citrix. Citrix Virtual Apps and Desktops and Citrix DaaS support OpenShift Virtualization as a hosting platform. Admins keep using Citrix Studio and Machine Creation Services (Red Hat blog), through a Citrix-managed OpenShift plug-in (Citrix docs).
  • VDI with Omnissa Horizon. Omnissa and Red Hat announced a collaboration in December 2025 to run Horizon 8 on OpenShift Virtualization. At announcement, manual pools and farms already worked, with a fully integrated solution promised later (Omnissa). Check Omnissa’s current support matrix before planning automated pools.
  • HPE Morpheus. HPE Morpheus Enterprise can onboard OpenShift clusters and provision, monitor and lifecycle-manage OpenShift Virtualization VMs from the same control plane as its other clouds (HPE Community, Red Hat Ecosystem Catalog). Note this is Morpheus Enterprise, not the cheaper VM Essentials tier.

These partners document support for OpenShift Virtualization, which is the same code OVE runs. If you buy OVE specifically, confirm with the vendor that the edition is covered by their support statement.

Backup: your existing investment probably carries over

Most major backup vendors now protect OpenShift Virtualization VMs, so switching hypervisor rarely means switching backup product. Red Hat lists Cohesity, Commvault, Dell PowerProtect, Rubrik, Veritas NetBackup, Portworx, Trilio, Storware and CloudCasa among its backup partners (Red Hat blog).

VendorProductStatusWorth knowing
VeeamBackup & Replication 13.1+Supported, new in 2026Native support via a KubeVirt plug-in, same console, no new license; can restore other hypervisors’ backups onto OpenShift and back (Veeam)
VeeamKasten for KubernetesSupportedFirst backup partner for OpenShift Virtualization (Veeam); v9 uses changed block tracking for incrementals (Veeam Community)
CommvaultCommvault CloudSupportedVMs and containers under one policy, file-level restore (Commvault)
CohesityDataProtect, NetBackupSupportedSame workflows as ESXi and Hyper-V sources (Cohesity)
RubrikRubrik Security CloudSupportedIncrementals, immutable backups, threat hunting in backups (Rubrik)
Red HatOADP (included, you install it)SupportedNo extra cost, Velero-based. You install the operator and configure a backup location. Snapshots alone stay on the cluster’s own storage; the Data Mover copies VM disks to S3-compatible object storage, which is what protects you if that storage fails. YAML-driven, no console of its own (OKD docs, NetApp docs)

The Veeam news matters most for this audience: since version 13.1, Backup & Replication protects OpenShift VMs from the console your team already uses, and Veeam lists it as a verified integration (Veeam). It is also a second migration path: back up on VMware, restore on OpenShift.

Two honest caveats. First, these products back up through Kubernetes and your storage’s CSI snapshots, so your storage choice must support snapshots and you should verify it against the backup vendor’s matrix. Second, install the QEMU guest agent in VMs that need consistent online backups, so snapshots can freeze the guest file system (Red Hat blog). That helps, but it is not a guarantee of application-consistent backups, so check each vendor’s documented limitations against your VM types before you rely on it. Your existing license may also need a different workload type or add-on, so ask your vendor before assuming it is free.

Where it really is harder

OVE is not a beast, but it is not Proxmox either. These are the places a VMware admin will spend real time:

  • Storage is a design decision, not a datastore. VMs live on persistent volumes from a storage class. Live migration needs shared storage that every node can reach, via OpenShift Data Foundation or a certified CSI driver from your SAN or NAS vendor. Check whether ODF is in your quote; do not assume it is. If you liked vSAN, look at Portworx: it pools the servers’ local disks into shared storage inside the cluster, with a ready-made storage class that makes VM disks live-migratable (Portworx). It also adds synchronous and asynchronous DR for VMs (Efficiently Connected) and supports the two-node-plus-arbiter layout (Red Hat blog). It is a separately paid product and one more vendor in the stack.
  • The VDDK image. For VMware sources, MTV uses VMware’s Virtual Disk Development Kit. You download it from Broadcom, build a container image and push it to your own registry, and Red Hat warns that putting it in a public registry may break VMware’s license (Red Hat docs). Warm migrations are refused without it (Red Hat KB).
  • DNS and networking up front. API and wildcard DNS records before install, and VLAN or bridge mappings for VM networks afterwards. Nothing exotic, but more prerequisites than an ESXi host.
  • New vocabulary. Namespaces instead of folders, storage classes instead of datastores, operators instead of plug-ins. You can drive most of it from the console, but troubleshooting eventually means oc on the command line.
  • Hyper-V migration is not production-ready yet. If your source is Hyper-V, MTV support is a Technology Preview today.
  • No permanent free tier. Proxmox is free to run in production without a subscription (Petronella). OVE is paid after the trial, although eligible existing Red Hat customers can currently get a year of fully supported OVE at no cost through Level Up. That is a promotion, not a free tier.

None of these are blockers, and a lab is the right place to meet them. Getting comfortable in a lab takes days. Being ready to design, upgrade, troubleshoot and recover a production OpenShift environment takes considerably longer, so budget for training, or a partner, for your first production cluster.

So which should you pick?

Pick by what you need in three years, not by which installer feels most like ESXi today.

  • Choose Proxmox if you want the lowest cost, a familiar model, and your team is happy with community or Proxmox-direct support. It is a solid platform, not a compromise.
  • Choose Hyper-V if you are already a Microsoft shop with Datacenter licenses and mostly Windows guests.
  • Put OVE on the shortlist if you want Red Hat’s support and partner ecosystem, migrate from vSphere at scale with warm migration, already use Red Hat tools like Ansible, run Citrix or Horizon VDI, manage several platforms from HPE Morpheus, want to experiment with AI agents against your infrastructure, or expect to run containers next to your VMs within a few years.

The best way to judge is a lab. A single spare server and the Assisted Installer are enough for a first look on a self-supported evaluation; for a proper proof of concept, use the official 60-day OVE trial on three nodes. Either way, migrate a couple of test VMs with MTV and decide for yourself whether OpenShift is really the beast its reputation says.

Sources

Leave a Reply

This site uses Akismet to reduce spam. Learn how your comment data is processed.

More posts

All posts →