Why I Still Run Hyper-V in 2019

"Just move everything to the cloud." I hear this at least once a month. It's usually said by someone who's never managed an on-prem environment and doesn't understand why some workloads stay where they are.

In 2019, Hyper-V is still the right answer for a lot of workloads. Not because cloud is bad. Because not everything belongs there.

The workloads that stay on-prem

Some workloads are on-prem for technical reasons. Legacy applications with latency requirements that cloud can't meet. Systems with compliance requirements that mandate physical control. Factory floor systems that can't tolerate internet dependency.

But more often, workloads stay on-prem for economic reasons. A SQL Server cluster running 24/7 on reserved instances costs significantly less on-prem over five years than in Azure, even accounting for power, cooling, and admin time. The cloud premium for always-on, predictable workloads doesn't always make financial sense.

I'm not anti-cloud. I run plenty in Azure. But I'm pro-arithmetic. If the numbers say on-prem is cheaper for a given workload over its lifecycle, on-prem is where it stays. Ideology doesn't pay the bills.

Hyper-V vs. VMware in 2019

VMware has the enterprise features, the ecosystem, and the mindshare. Hyper-V has integration with the Microsoft stack and a price tag that's hard to beat when you're already licensed for Windows Server Datacenter.

For a pure Windows environment, Hyper-V with SCVMM is completely capable. Live migration, failover clustering, replica for DR — it's all there and it works. The management experience isn't as polished as vSphere, but it's functional, and the PowerShell automation is excellent.

Where VMware wins: mixed environments with significant Linux workloads, NSX for micro-segmentation, and third-party ecosystem integrations. If you're running 50% Linux, VMware is the better choice. If you're 90% Windows, Hyper-V does everything you need for less money.

The hybrid reality

The most practical architecture in 2019 is hybrid: Hyper-V on-prem for steady-state workloads, Azure for burst capacity and new projects. Azure Site Recovery for DR. Azure Backup for off-site retention. Azure AD Connect for identity. Not "all cloud" or "all on-prem" — both, where each makes sense.

I run Hyper-V clusters for the ERP, the file servers, the domain controllers. I run Azure for customer-facing web applications, dev/test environments, and anything that benefits from elastic scale. The two environments are connected, managed with the same tools, and treated as one infrastructure platform.

This isn't exciting. It's not a conference talk topic. But it works, it's cost-effective, and it's more common than the cloud-only narrative would have you believe. Most enterprises are hybrid. They'll be hybrid for years. That's fine. The goal isn't to be pure. The goal is to be effective.