Every list of VMware alternatives is written by someone selling one of them. This one is too. So here is the shortlist with the case against each option, including ours.
The Broadcom renewal has pushed a lot of infrastructure teams into a decision they were not planning to make this year. Most of the advice they find compares hypervisors, which is the least interesting part of the problem.
What Are You Actually Replacing?
A vSphere cluster is four things at the same time:
- Pooled capacity across several hosts, so a VM is not tied to one machine
- Live migration, so hardware can be patched without downtime
- Shared storage that every host can see
- One console, and a decade of operational habits built on it
Every alternative below should be judged on those four, not on whether it runs KVM or Hyper-V. A hypervisor swap is a weekend. Losing live migration is a change to how your team patches hardware for the next five years.
Option 1: Stay and Pay
The option nobody writing one of these lists wants you to pick, and the correct one more often than the internet suggests.
Migration is not free. The real bill is an assessment, a parallel environment running alongside production, a cutover window, re-testing every backup job and DR runbook that assumed the old platform, and retraining the people who operate it. For a small estate — a handful of hosts and a few dozen VMs — that project can cost more than several years of the increase it is meant to avoid.
The honest threshold: work out the three-year cost of the migration project, including the staff time nobody puts on the invoice, and compare it to three years of the renewal difference. If migration loses, stay, and spend the time on something that earns money instead.
Option 2: Hyper-V
The closest operational analogue if your estate is mostly Windows. Failover Clustering and Cluster Shared Volumes cover pooled capacity and live migration, the management story is familiar, and the people who run your Windows servers already understand the model.
Where it is wrong: you have moved the licensing conversation from one large vendor to another, and Datacenter edition is priced per core. The third-party ecosystem — backup, monitoring, automation — is real but thinner than VMware’s. If your estate is mostly Linux, this is a strange destination.
Option 3: Nutanix
The like-for-like enterprise answer. Hyperconverged, its own hypervisor, a management plane that experienced vSphere admins get on with quickly, and a vendor who will turn up to the meeting.
Where it is wrong: you are swapping one vendor’s licensing model for another’s. The renewal conversation comes back in three years with a different logo on it. If what drove this project was being exposed to a single vendor’s pricing decisions, this solves the symptom rather than the condition.
Option 4: Proxmox
Proxmox has taken most of the public conversation since the Broadcom changes, and it deserves it — it is a capable platform with a real clustering story and no licence gate on the features.
We sell Proxmox, and we will still tell you what it is not. Ours runs on a single bare metal machine with a ZFS mirror. One node. That makes it an excellent lab, a dev and staging environment, or a replacement for a single ESXi host that was never clustered in the first place. It is not a clustered vSphere replacement, and we do not sell it as one. If you need capacity pooled across hosts and VMs that survive a machine failing, a single node does not give you that, whatever the hypervisor underneath.
We also do not resell Proxmox subscriptions. We provide the image; if you want the enterprise repository and vendor support you buy that from Proxmox directly. That is a cheaper arrangement for you and an honest one for us.
Option 5: A Reserved Pool at a Provider
This is the model that actually matches what you are leaving, and the one most buyers do not know is available, because the phrase for it has been taken.
At a hyperscaler, “VPC” means a software-defined network boundary. Private address space, subnets, route tables — and your virtual machines still run on hypervisors shared with other customers. It is free, because it is a configuration object rather than a resource. It is about packets, not contention.
A reserved pool is the other meaning: a block of vCPU, memory and storage held for you, inside which you build whatever you like. That is the shape of a vSphere cluster, and it is the thing the hyperscalers do not sell below very large commitments.
Ours is the Virtual Data Centre. VDC Starter is $499 a month flat — 20 vCPU, 64 GB of RAM, 2 TB of storage, 2 TB of traffic, four public IPs and a virtual router, in Dubai DX1, Fujairah FJ1 or Vienna VIE2. No per-VM charge: build twelve small machines or three large ones out of the same allocation and the invoice does not move. The private network, security groups and router are self-managed through Cloud Control.
Where it is wrong: if you intend to keep operating the estate exactly as you do today, with your own tooling against your own hardware, a reserved pool on a shared platform is not that. Managed Private Cloud on dedicated hardware is the closer fit — but that is a project with a design phase, not a plan you order on a Tuesday.
The Comparison, Honestly
| Option | Gives you | Where it is the wrong answer |
|---|---|---|
| Stay and pay | Zero project risk | Large estates where the increase compounds |
| Hyper-V | Clustering, familiar to Windows teams | Linux-heavy estates; licensing moves, it does not vanish |
| Nutanix | Closest like-for-like, vendor support | If vendor exposure was the actual problem |
| Proxmox (single node) | Labs, dev and staging, one-host replacements | Anything needing capacity pooled across hosts |
| Reserved pool at a provider | The cluster model without the hardware | Teams who must keep their own tooling and hardware |
How to Choose
- Count the VMs that genuinely need live migration. If the answer is “none, we patch in a maintenance window”, half the shortlist opens up.
- Price the project, not just the platform. Three years of renewal increase against the full cost of moving, staff time included.
- Decide whether you are escaping a price or a dependency. They lead to different destinations.
- Check what your backup and DR tooling supports before you choose, not after. This is where migrations actually fail.
- Separate production from everything else. Labs and staging can move first, cheaply, and tell you what you did not know.
What We Do Not Do
We do not sell clustered Proxmox. We do not resell Proxmox subscriptions. Managed Private Cloud is a design engagement before it is a product, so it is not something you can order and have running this afternoon. And if your estate is small and your renewal is modest, we will tell you to stay where you are — that conversation costs us a sale and saves you a project.
This is part one of a series on leaving VMware. Next: what Broadcom actually changed, and how to work out your own number.
If you want the short version for your own estate — tell us what you are running and we will tell you which of the five it points at, including when it points at none of ours.
