Alternative to VMware: what to assess before changing infrastructure

Tecnico IT in un data center: Nutanix come alternativa a VMware

The renewal quote arrives in September and looks nothing like last year’s. Nobody had planned to touch the virtualisation layer: the environment worked, backups ran, the cluster still had a few years of service ahead. Then the IT manager takes the document to the board and the conversation changes. This is how many companies start looking for an alternative to VMware.

The motivation is legitimate. As a basis for a decision, though, it falls short, because the licence is only one of the items at stake and rarely the one that weighs most over the medium term.

People looking for an alternative to VMware assess the hypervisor, but the problem lies around it

In a traditional infrastructure servers run the workloads and data sits on separate storage, often a SAN. They are distinct components, with distinct consoles, distinct update cycles and distinct skills to keep them running. While everything works, that complexity stays invisible. It shows up at the first failure, when a firmware needs updating, when more capacity is needed and the bottleneck turns out to be storage rather than servers, or when the person who knew the configuration leaves the company.

Hyperconvergence changes the pattern. Each node provides compute and local storage, and the software pools the resources and presents them as a single cluster: virtualisation, storage, networking and operations converge under one control plane. Nutanix should be read in this light, as a different way of integrating components rather than a replacement hypervisor. The value, when there is any, comes from integration more than from swapping a single component.

The difference shows up in a working day more than in a datasheet. Whoever today opens three consoles to trace why a virtual machine is slow opens one; whoever coordinated two vendors for a firmware update follows a single procedure. Sales presentations rarely mention this; the IT team notices it within the first month.

Seventy-eight virtual machines and sizing built on caution

In a migration project we worked on, the starting infrastructure was a three-node cluster with seventy-eight virtual machines and one hundred and thirty-two virtual cores assigned in total. The analysis brought out a detail we see often: a great many machines had a single vCPU, a sign of real consumption far lower than the configuration suggested. Memory told the same story, installed generously and used at less than half. Resources had been assigned out of caution, not from a measurement.

The sizing proposed to replace that cluster was three single-socket nodes, each with sixteen cores and 384 GB of memory: a system built on usage data instead of a copy of the existing one. In our internal technical review we discussed whether to raise the core count or the processor frequency, and chose frequency, because the data showed fewer cores were needed, not more. It is the kind of decision that a quote mirroring the existing setup does not even allow you to frame. The target hyperconverged platform carried the rest of the project: migration in groups of machines during agreed windows, replication between nodes active from day one, and a single console for compute and storage that the internal team learned to read during hands-on support.

Signs that the infrastructure has reached its limit

It is rarely a failure. More often it is a sum of minor symptoms, and these are the ones we meet most often during assessments.

  • Backup windows stretch until they nearly reach working hours.
  • A vendor proposes extra hardware before checking how the existing hardware is used.
  • Updates are postponed because nobody wants to touch that configuration.
  • Nobody in the company can say how many resources are actually in use.
  • The restore procedure has never been tested, or is not documented.

When these signs appear together, the issue is no longer the licence: routine maintenance is turning into an operational risk.

Moving to an alternative to VMware becomes the natural choice when conditions like these recur: end-of-life hardware that has to be replaced anyway, a particularly costly licence renewal, the need to revisit backup and business continuity, or a small IT team struggling to maintain a fragmented architecture. In those cases the migration coincides with work that is needed anyway, tackled in one go instead of in separate interventions, and the signs listed above indicate that the moment has come.

Alternative to VMware: what to compare beyond the licence

Anyone assessing an alternative to VMware compares at least five items in writing, and the first is the one that matters least.

  • Licences, the starting point of the reasoning and rarely the deciding factor.
  • Hardware, bearing in mind that sizing based on real consumption can reduce the number of nodes compared with today.
  • Day-to-day management time, which tends to fall in an integrated system, provided the team learns a new console.
  • Migration, whose weight depends on the number of virtual machines, how critical they are and the downtime windows available.
  • In-house skills, the most underestimated variable: a platform nobody in the company can run shifts the dependence on the vendor instead of reducing it.
Item Renewing the existing setup Moving to hyperconvergence
Licences Fee renegotiated at every renewal Model tied to the cluster nodes
Hardware Servers and SAN replaced separately at end of life Nodes sized on actual consumption
Day-to-day management Separate consoles for compute, storage and network A single control plane
Business continuity Tied to restoring the shared storage Replication between nodes built into the design
Skills Separate profiles for each component One platform to learn

To the five items it is worth adding three questions for whoever proposes the change: over what period consumption was measured, which application dependencies have been mapped, and what happens if a workload does not behave as expected halfway through the migration. The answers separate a project from an order.

How to move to an alternative to VMware without stopping the business

You get there by measuring real consumption, sizing on the data and migrating in groups of machines, from the least critical to the essential ones, during agreed windows. The sequence allows few shortcuts. Measurement covers a representative period, a few weeks rather than one day. Sizing built on that data follows. Then a first group of non-critical machines is chosen, to validate process and timing on workloads that can wait. Only after that come the systems that cannot stop, with agreed windows and a rollback plan defined before starting.

Surprises rarely come from moving the virtual machines; they come from dependencies: applications tied to fixed IP addresses, licences bound to the hardware serial number, integrations with external systems that nobody ever documented. They emerge in almost every project, and the right phase to find them is the analysis.

An infrastructure migration rarely stays confined to the hypervisor. When part of the workloads moves off site, the topic becomes managed cloud, while the day-to-day management of what stays on premise falls within fixed-fee managed IT.

At Aesir Srl we are a Nutanix Professional Partner and approach these projects by measuring workloads rather than starting from the price list. We work on projects across Italy and, on specific request, in Switzerland, the United Kingdom and the United States, with our own Tier IV data centres in the European Union and replication between sites, within a perimeter certified ISO 9001, 27001, 27017 and 27018. For anyone assessing a platform change the practical consequence is one: migration, target infrastructure and ongoing management have a single owner, because we act as MSP and MSSP under the same contract. Platform specifications are documented on the official Nutanix website.

Frequently asked questions

Can you migrate without stopping production?

Yes, in most cases: many machines move with short, schedulable interruptions, while critical applications get an agreed window. Dates are set during the analysis, workload by workload.

Does moving to an alternative to VMware require new hardware?

Usually yes, because hyperconvergence is based on nodes that combine compute and storage. The number of nodes, however, depends on real consumption and can turn out lower than today.

Can the in-house team manage it?

Yes, with a period of hands-on support. The single console simplifies daily tasks; training focuses mainly on backup and restore procedures.

How is real resource consumption measured?

With analysis tools that collect data over a few weeks and report actual processor, memory and storage usage for each virtual machine. The same data then forms the basis for sizing.

Let’s talk

Before choosing an alternative to VMware it pays to get hold of one number: how many resources are actually used compared with those assigned. The current infrastructure can provide it, and it often changes the terms of the decision, because it moves the discussion from the price of a licence to the sizing of a system.

If you would like to explore the topic or assess your company’s situation, fill in the form at the bottom of the page or write to support@aesir-tech.it: we will book a free consultation and start from your numbers.

Vuoi parlarne applicato alla tua azienda?

Ti rispondiamo con l’esperienza di progetto, non con una brochure.

Request a call