In December 2023, weeks after Broadcom closed its acquisition of VMware, the company announced the change that the historical pattern had suggested was coming: VMware is moving to a subscription-only model, and perpetual licenses — the buy-once, own-forever model that the bulk of the installed base relied on — are being discontinued in favor of subscription bundles. For organizations that had built private-cloud and datacenter infrastructure on VMware over the past decade, this turned an abstract acquisition-risk question into a concrete budget and architecture question. January 2024 is, accordingly, peak evaluation season: across the enterprise and public sector, infrastructure teams are pricing renewals under the new model and, for the first time in years, seriously surveying the alternatives.
This piece is a survey of those alternatives. It is not a recommendation to migrate — for many organizations, particularly large VMware estates, staying and absorbing the new commercial terms will be the rational choice once the numbers are run. But every organization affected should now understand the landscape, because the credible option to leave is itself the basis for any negotiation, and because some workloads will genuinely be better placed elsewhere.
Framing the Decision
Before surveying tools, it helps to be clear about what migration off VMware actually involves. A VMware estate is rarely just hypervisors. It is vSphere for virtualization, vCenter for management, often NSX for networking and vSAN for storage, plus a surrounding ecosystem of backup, monitoring, and automation tooling that integrates with the VMware APIs. The cost of migration is dominated not by moving virtual machines — that part is increasingly well-tooled — but by reproducing the operational capabilities the surrounding stack provided, and by retraining staff whose skills and certifications are VMware-specific.
That means the right unit of analysis is the workload and its operational requirements, not “the datacenter” as a monolith. Some workloads are simple VMs that will move readily. Others are deeply entangled with VMware-specific networking, storage, or management features and will be far harder to relocate.
The Open-Source Options
Proxmox VE has emerged as the most-discussed near-term alternative for organizations seeking a like-for-like virtualization platform. Built on KVM and LXC with an integrated web management interface, Proxmox offers VM and container management, clustering, and built-in backup, under an open-source model with optional paid support subscriptions. Its appeal in the current moment is pragmatic: it is mature, it is relatively approachable for teams with VMware operational habits, and its commercial model is far less expensive than the subscription bundles now on offer. Its limits show at the largest scales and in the most demanding software-defined networking and storage scenarios, where the VMware stack’s depth is hard to match.
OpenStack is the most capable open-source private-cloud platform and the natural destination for organizations that want a full cloud — self-service provisioning, multi-tenancy, software-defined networking and storage, and an API surface comparable to a public cloud — rather than just a hypervisor. Its historical drawback is operational complexity: standing up and running OpenStack well requires genuine cloud-operations capability. That barrier has fallen considerably thanks to mature distribution packaging from commercial vendors, and OpenStack is deployed at scale in telecommunications, research, and government environments. For a large organization with the operational depth to run it, OpenStack is the option that most fully replaces what a sophisticated VMware private cloud provided.
OpenNebula occupies a deliberately lighter-weight niche. Designed for VM and container orchestration without the full OpenStack service catalog, it suits organizations that want straightforward private-cloud management — particularly across distributed or edge sites — without OpenStack’s operational footprint. Its lineage is worth noting on a site with this heritage: OpenNebula grew out of distributed-systems and grid-computing research into resource scheduling and the federation of computing resources across sites, and that research DNA shows in its focus on managing capacity across multiple, possibly heterogeneous, locations. For organizations whose VMware use was essentially “manage a pool of VMs efficiently,” OpenNebula is often a closer fit than the heavier platforms.
The Commercial Alternatives
Nutanix offers a hyperconverged infrastructure platform with its own AHV hypervisor (also KVM-based) and an integrated management layer, positioned explicitly as a VMware alternative. It is a commercial product with commercial pricing, so it does not escape the subscription model in principle — but it gives organizations a single-vendor, supported migration target with a different commercial relationship and a strong story for the hyperconverged use case.
Microsoft Hyper-V is the obvious option for organizations already invested in the Microsoft ecosystem. As a virtualization platform bundled with Windows Server and integrated with the broader Microsoft management stack, it can be a low-friction destination for Windows-centric estates, and its licensing may be partly absorbed by existing Microsoft agreements.
Public-cloud migration is the alternative that sidesteps private virtualization entirely. For some workloads, the VMware licensing change is the prompt that finally tips the analysis toward rehosting in a public cloud — either as lift-and-shift VMs or, over a longer horizon, as re-architected cloud-native services. This is not free of lock-in concerns of its own, but for workloads that suit it, the licensing disruption is a natural decision point.
Practical Guidance for the Evaluation
For teams running their numbers this season, a few principles keep the analysis grounded:
- Run the renewal numbers first. Before evaluating migration, get a clear figure for what staying costs under the subscription model. Migration is justified only relative to that baseline, and for some estates the baseline will be tolerable.
- Segment the estate. Identify the workloads that are simple VMs (easy to move) versus those entangled with VMware-specific networking, storage, or management features (hard to move). A partial migration of the easy workloads can reduce the licensed footprint without a wholesale platform change.
- Cost the operational transition honestly. The headline savings of an open-source platform are offset by the cost of building or buying the operational capability and support that VMware provided. Price the staffing, training, and support realistically — an underutilized or under-operated alternative can cost more than the licensing it replaced.
- Pilot before committing. Stand up the candidate platform with a representative non-critical workload and measure the real operational experience before betting a production estate on it.
- Preserve negotiating leverage. Even for organizations that ultimately stay, the credible, well-understood option to leave is what makes the renewal negotiation a negotiation rather than a notification.
The licensing change is recent and the migration wave it is provoking is just beginning. The decisions organizations make this season will play out over years. The teams that fare best will be those that treat the moment not as a panic but as the long-overdue occasion to understand, in detail, what their infrastructure depends on and what the alternatives genuinely cost.
Further Reading
- Open Infrastructure Foundation — governs OpenStack and publishes deployment surveys relevant to private-cloud evaluation.
- Linux Foundation — neutral-governance home for KVM and related open virtualization projects underpinning several alternatives.