1. Why many businesses are reviewing VMware
For years, server virtualisation was a technical matter that was decided once and then forgotten. That is no longer the case. VMware's licensing models have changed, with moves towards subscriptions and new minimum thresholds, and many businesses have found that the renewal cost is no longer what they planned for when they built their infrastructure. The problem is not only price: it is dependence on a single supplier for a component that underpins all of the company's systems.
This is why the question has shifted. It is no longer just which platform is more convenient, but how easy it would be to change course if the terms deteriorate. Open platforms, such as Proxmox VE based on KVM, offer a more negotiable way out, but they require skills and a plan. Before deciding, it is worth understanding what you are actually protecting: continuity, predictable costs, control of data or freedom to choose suppliers.
2. What KVM and Proxmox VE are, in plain terms
KVM is the virtualisation technology built into the Linux kernel. It allows several virtual machines to run on the same server, each with its own operating system, making efficient use of the processor. On its own it is an engine: it includes no graphical interface, backup, clustering or user management. Anyone who uses it directly must build these tools from separate components and maintain them over time.
Proxmox VE is a platform that combines KVM and Linux containers in a single management console, with clustering, high availability, integrated backup and data replication between nodes. It is a popular choice for those who want a complete, open solution without licences for each virtual machine. Commercial support is available as a separate subscription, and its level should be chosen according to the criticality of the systems. The software may be free, but running it is not.
3. Comparing the possible routes
There is no universally better platform. The choice depends on the number of servers, the skills of the team, the applications that need to run and the level of support required. The table compares the four most common routes on practical criteria. The assessments are qualitative and should be verified against your own case. Use them as a framework for discussion with the supplier, not as a verdict.
A common route is to keep VMware on the most critical systems and gradually move the less sensitive workloads. This reduces risk, but it means running two platforms for a longer period. Ask the supplier to explain the total cost of both routes, including training, support and staff time, not just the licence price. The most useful comparison is between two complete scenarios, over three or five years.
| Option | Strengths | Limits to consider | When to choose it |
|---|---|---|---|
| VMware vSphere | Mature ecosystem, established tools, a large pool of skills on the market | Costs and licensing terms subject to change, dependence on one supplier | Large environments with stable contracts and an already trained team |
| Proxmox VE | Complete and open platform, integrated clustering and backup | Requires Linux skills and disciplined maintenance | SMEs and mid-sized businesses with a technical team or a support provider |
| KVM with your own tools | Maximum control and zero licence costs | Every management component must be built and maintained | Expert team with very specific requirements |
| Managed public cloud | No hardware management, elastic scalability | Dependence on the provider and costs that vary over time | Variable workloads or no internal data centre |
4. How to plan the migration
A successful migration does not start with installing the new hypervisor, but with the inventory. You need to know which virtual machines exist, which applications run on each, what dependencies they have between them and who is responsible for each. Many surprises come from a forgotten server that a nightly process still uses, or from an application tied to a driver specific to the previous platform. For this reason the inventory should be carried out with the technicians who manage the systems today, not only from the documentation.
Each machine should then be classified by criticality and by the maintenance window available. Non-critical machines can be moved first, to learn the process. The most important ones require a full test, with restore verified from backup and a rollback plan in case anything goes wrong. The migration should be thought of as a series of reversible steps, not as a single cutover day.
- Inventory virtual machines, applications, dependencies and owners.
- Classify each system by criticality and by acceptable maintenance window.
- Verify that existing backups really restore, in a test environment.
- Build the target environment with storage, network and cluster sized to the real workload.
- Migrate the least critical systems first and document timings, problems and fixes.
- Run a full restore test after each group of machines has been moved.
- Keep the previous platform active for an agreed period as a rollback route.
5. Technical risks and common mistakes
The first mistake is underestimating storage and networking. An open platform is flexible, but performance depends on disk quality, volume configuration and VLANs. An environment designed too lightly becomes slow precisely at peak times, and the blame falls on the new software even when the problem is sizing. A load test before the cutover avoids most of these surprises.
The second mistake is thinking that the migration ends at go-live. Backup procedures, monitoring, access rights and documentation still need updating. The third is changing platform without an internal owner or a supplier with whom maintenance is agreed. When the systems engineer who built everything leaves the company, an undocumented platform becomes a new dependency risk.
6. What to ask the supplier before deciding
If you do not have a sufficient internal team, the IT services provider becomes part of the decision. Ask which platform they propose and why, what technical references they have on similar migrations and how they handle security updates. Also ask how the infrastructure will be documented, who will hold the administrative credentials and how they will be safeguarded. Ask too what response time is committed to in the event of an overnight failure.
A good supplier should also tell you what it will not do. If it proposes a full migration in a few days without testing, or promises results without knowing your inventory, that is a warning sign to investigate. Ask for a written plan with phases, success criteria and separate costs for the migration and for ordinary management. If the supplier cannot separate them, it is not working on your architecture.
7. How to start: the first concrete step
The first step is to understand the current dependency. Ask your current licence provider for the complete picture of contracts expiring, the services included and the renewal conditions. Alongside this, prepare an inventory of servers and workloads, with the criticality of each. With these two documents you can compare the routes without deciding under pressure. If this analysis takes more than a few days of internal work, that is already a useful signal.
Then set up a test environment with a subset of non-critical machines. Within a few weeks you will see what is really needed: skills, timescales, backup tools, maintenance procedures. Only then does it make sense to evaluate the full plan and its real cost. The pilot gives you real data on timings, skills and problems, which no price list can give you.
Frequently Asked Questions
Can I migrate from VMware to Proxmox without stopping services?
In some cases yes, with a live migration for compatible machines and a maintenance window for the others. It depends on the applications, the storage and the tolerance for interruption. The safest method is to move the least critical systems first and plan a short intervention with verified restore for the important ones.
Is Proxmox suitable for a business with a few dozen servers?
It can be, provided the team or the supplier has the skills to manage clusters, storage and backup. Size is not the only criterion. What matters is the availability of people who can respond when something fails, and a regular update process.
Is open-source software really more secure?
Security depends on maintenance, not on the licence. An open product can be updated and checked by many people, but if nobody applies the fixes it remains vulnerable. Ask how patches are received and applied, and who responds in the event of an incident, both for a commercial platform and for an open one.
Can VMware and the new platform coexist during the transition?
Yes, and it is a common route. The two platforms can coexist for a period, with critical systems still on VMware and less important ones already migrated. However, backup, monitoring and access need to be managed separately, and a clear schedule for decommissioning is required.
How much does the licence matter in the total cost?
It matters a great deal, but it is not the only cost. Migration time, training, support, backup tools and routine maintenance must all be considered. An honest comparison puts the total cost over several years on the table, not just the price of the first year's licence.