VMware to Microsoft: Customer Stories

This one skips the theory. Daniel, a Microsoft MVP in Australia who specializes in Azure hybrid solutions, came back to walk through three real migrations his team ran: what went sideways, and what the customers got out of it.

Daniel runs a managed service provider that’s been on Hyper-V since Server 2008. These are clusters his team built and patched, one of them converted after a couple of glasses of wine at dinner.

A manufacturing rollup with a PC on a shelf

The first story is a manufacturing company in Australia that grew through acquisition and ended up with mismatched infrastructure across its factories. Operations teams, not IT, had been buying the tech. A vendor would show up with a new labeling or scanning system, drop off a basic VMware box, and walk away. One factory ran a critical labeling system worth hundreds of thousands of dollars a day in production off a $200 desktop PC sitting on a cabinet.

Daniel’s team consolidated it onto Hyper-V with Storage Spaces Direct. Three sites got two-node clusters for high availability, two smaller remote sites got single-node Hyper-V hosts, and everything moved onto one AD domain, replacing a mess of local accounts and separate vCenter logins.

The factories ran 24 hours a day, six days a week, so uptime mattered. Hyper-V Replica handled disaster recovery between factory sites, and head office workloads replicated into Azure via Azure Site Recovery. Migration tools were nothing exotic: Microsoft Virtual Machine Converter for the older VMware 5.5 environments, Disk2vhd for the rest.

Cost was the whole point. A VMware quote for two nodes plus an iSCSI SAN ran around $150,000. Daniel’s team built it on commodity 1U servers from HP and Lenovo with local SSDs instead. The customer has since upgraded those clusters to Azure Stack HCI and is still running them, with a check-in message from the customer’s engineer just weeks before this episode aired.

A pet food company and a four-hour report

The second story is another acquisition-driven FMCG company, this one in pet food. They bought a smaller company running its own VMware cluster on iSCSI storage: a Citrix farm and a handful of SQL and app servers, about 30 workloads total.

The approach was two-stage. First, ship hardware onsite and convert VMware to Hyper-V locally, so the conversion runs at local bus speed instead of crawling across a WAN link. Then use Hyper-V Replica to move everything into the main data center over a week or two, throttled so it wouldn’t saturate the link. The final cutover happened during a scheduled maintenance window. End users experienced it as a reboot.

The performance jump was the real story here. An overnight payroll and reporting job that used to take about four hours finished in five to ten minutes after the move to Storage Spaces Direct. The first time it ran, the customer thought something had broken and called in a panic, because the report was sitting there finished before anyone had left for the day.

It wasn’t a completely clean night. A production manager wasn’t told a workload was going offline for half an hour during the conversion, and Daniel’s phone started ringing with people who couldn’t print shipping labels. A communication gap, not a technology problem, but a reminder that the people running the floor need the same heads-up as the people running the cluster.

Critical infrastructure with zero room for error

The third story is a government critical infrastructure customer in Western Australia running about 115 workloads, a mix of line-of-business apps and operational technology, on an aging VMware 6.5 cluster that was out of warranty and hitting performance limits. They wanted a fully managed Azure Stack HCI build with NVMe in front of spinning disk, and someone else carrying the operational load.

The migration hit a real obstacle. The plan was to use System Center Virtual Machine Manager for the conversion, but it kept failing on FQDN and domain credential issues Daniel’s team couldn’t run down despite years of SCVMM experience. They pivoted: about 60% of workloads moved with StarWind V2V during tolerant maintenance windows, and the 40% that couldn’t tolerate downtime moved with Carbonite. A 10 Gb cross-connect between data centers, backed by a secondary Megaport circuit, kept large VM conversions to 15 to 20 minutes even around 120 GB per VM.

The performance gain echoed the pet food story: an app owner called in convinced something was broken, because the NVMe-backed cluster, benchmarked at roughly 2.8 million IOPS across five nodes, made a familiar application feel instant.

Two rough edges stood out. One was a genuine bug: removing SCVMM left a leftover DHCP filter driver on the virtual switch, causing intermittent VM-to-VM connectivity drops that took Microsoft a few months to fix. A similar issue showed up again later with no clear root cause. The fix both times was a live migration to force the two VMs onto the same node. Now the team drains and tests a node a day or two ahead of any patch cycle instead of letting Failover Clustering disperse VMs automatically.

One more detail worth noting: the customer got read access to Windows Admin Center through the Azure portal, with RBAC, MFA, and no VPN required. They loved the concept until the VM inventory view, which runs on WMI polling, slowed down past 60 or so workloads. It’s improved since, but it’s a real limit to know about before leaning on Windows Admin Center as a day-to-day tool at scale.

The pattern underneath all three

Every one of these was a managed environment, which flattens the skills-gap problem that scares people off Hyper-V. MSPs gravitate to Hyper-V because it lowers their cost of platform. Internal IT teams that aren’t watching every dollar as closely tend to stick with VMware longer. Cost pressure is what moves people, not a features chart.

Daniel also said the conversations have shifted since the Broadcom news broke. Customers used to ask whether Hyper-V would work. Now they’ve already decided to move and just want help planning it. He’s mid-engagement helping an energy sector customer convert about a dozen more clusters right now.

This post is part of the VMware to Microsoft podcast series.

Subscribe for more of these.