VMware to Microsoft: Who's Running Azure Stack HCI
John Miller, a 30-year technology consultant, on real Azure Stack HCI migrations: a 350-site automotive rollout, police departments ditching their SANs, and why AVD is moving back on-site.
Most of this podcast has been making the case that Azure Stack HCI and Hyper-V are real alternatives to VMware. Episode 11 answers a different question: who’s running it, and what happened when they got there. John Miller, a 30-year technology consultant who’s worked for Microsoft and several Microsoft partners, walked me through the projects he’s got in flight right now.
John specializes in data center and cloud technology now, specifically Azure and Azure Stack HCI. He spends a lot of his time running demos on Microsoft and Dell Azure Stack HCI environments on the 23H2 release, showing customers what the platform does before they commit to anything.
A 350-site automotive customer, skipping Hyper-V entirely
The biggest example John shared is a global automotive company with roughly 350 sites across four data centers. They’re moving straight from VMware to Azure Stack HCI, not Hyper-V, specifically because of the newer tooling and the Azure integration.
They’re new to Azure Stack HCI itself, but they already understand how clusters function from running their own data centers for years. That’s the profile John sees a lot: strong infrastructure fundamentals, just new to this specific stack. Cluster sizes across their sites range from two nodes up to six, because each site’s requirements are different. There’s no single template that fits all 350 of them.
Local government and utilities are done buying SANs
A second pattern John’s seeing is public sector customers: police departments, state governments. These are organizations with very little Azure footprint today, and they’re planning to stay on site rather than move to the cloud.
Two things are driving them. First, VMware licensing costs, even with Broadcom’s newer licensing paths like Azure VMware Solution. Second, and this is the part I found more interesting: the storage performance. These customers are seeing IOPS and throughput out of Storage Spaces Direct that’s comparable to their existing SAN and NAS gear, and some of them are getting rid of their SANs entirely and moving those VMs straight onto Azure Stack HCI clusters instead.
John was careful to note this isn’t just a government thing. Auto manufacturing shows up in the same conversations. The vertical matters less than the cost and value math, and right now that math favors moving off VMware.
Why some Azure-first customers are pulling workloads back on site
Some of John’s customers are already deep into Azure, running Azure Virtual Desktop or Kubernetes in the cloud, and they’re now bringing pieces of that back on premises onto Azure Stack HCI. I’ve seen the same thing, and the reasoning makes sense.
The driver for some is governance and compliance, while for others it is cost. One customer wanted their AVD environment hosted locally so remote contractors could VPN into the corporate network and reach a virtualized desktop that never left the building. Another has AKS clusters in Azure today and wants to run some of them on site for the same reason. Azure has a high compliance bar, but for these customers, physical control of the data still matters more than what a compliance certification can promise.
What a finished migration looked like
John described one completed project that used Azure Migrate, which was still in preview when they ran it. For the supported Linux and Windows guest operating systems, the migration went smoothly. No real hiccups.
The bigger win was how familiar it felt to the end users afterward. Deploying a VM through the Azure portal looks and feels the same whether the VM lands in Azure or on an on-site Azure Stack HCI cluster. You’re just picking a different location. Teams that already knew how to operate in the Azure portal didn’t have to relearn anything. John called the Azure Arc integration on top of that huge.
Across John’s customer base, he estimates about 80% of the workloads moving over are still plain VMs coming off VMware, Hyper-V, or Nutanix. Once that lands, customers start layering in AVD or AKS locally on the same cluster.
Handling the nervous customers
Not every prospect is sold immediately. John’s approach with a nervous customer is to find out specifically what they’re worried about, then demo it directly rather than arguing in the abstract. The most common questions are about storage: will it handle the IOPS and throughput they need, does it support deduplication. A lot of that hesitation is outdated thinking. The storage and hardware vendors in this space have solved those problems, and Storage Spaces Direct measuring up well against vSAN is one of the stronger data points he can show.
For anyone who wants to kick the tires before committing budget, John pointed to two low-commitment options: an all-in-one HCI jumpstart VM you can spin up and explore, and try-and-buy hardware programs some vendors offer so you can prove it out on real gear before you sign anything.
My take
Broadcom’s history isn’t buying a tool and making it better. It’s the opposite pattern: acquire, extract, let it stagnate. VMware was a genuine pillar of innovation for two decades, and Broadcom already sold off Horizon, its own VDI product, to another company. That should worry anyone still counting on VMware’s roadmap.
The customers John described aren’t edge cases. A 350-site automotive company and a state government aren’t early adopters chasing the shiny new thing. They’re organizations that need this to just work, and it is working for them. If you’re still on the fence, that’s the signal worth paying attention to.
This post is part of the VMware to Microsoft podcast series.
Subscribe for more of these.