Server Virtualization: Savings and Flexibility in Practice

Server virtualization means running multiple logically separated virtual machines (VMs) on a single physical device (host). Each VM has its own operating system, applications, and resources (CPU, RAM, disk). This utilizes a single physical device much more efficiently — in practice, we see an 8- to 15-fold increase in utilization.

TL;DR — Key Points

• Hyper-V is an excellent choice if you are already paying for Windows Server licenses
• Proxmox VE is a mature open-source alternative with no licensing costs
• VMware remains the standard for large environments, but has become significantly more expensive following the Broadcom acquisition
• Each VM has lower energy and space costs compared to a separate physical server

Why Virtualize

1. Consolidation: 5 application servers on 1 host.
2. High Availability: If a host fails, the VM automatically moves to another host (HA).
3. Snapshots and restore in minutes — indispensable for upgrades and incidents.
4. Isolated test environments without additional hardware.
5. Easier cloud migration: Transfer VMs to Azure, AWS, or Google Cloud.

Platform Comparison

How to Approach a Virtualization Project

1. Inventory — which servers are in production, what is the CPU, RAM, and disk consumption.
2. Consolidation analysis — rule of thumb: 3–4 typical VMs per core of a modern CPU.
3. Storage plan — SAN/NAS with redundancy, or local NVMe with replication?
4. Networking — 10 GbE for traffic between hosts and storage, separate VLAN for management.
5. Backup and DR strategy — snapshots are not a backup.
6. Migration — test VMs first, then production during a maintenance window.

Storage: A Common Pitfall

In practice, the most common bottleneck in a virtualized environment is storage. When purchasing a new host, do not skimp on disks — instead of SATA SSDs, go for NVMe or all-flash SAN. If you have demanding applications (SQL, ERP), monitor IOPS and latency, not just space. For small environments, Storage Spaces Direct (S2D) on Windows Server or Ceph in Proxmox is often the best choice.

High Availability (HA)

For HA, you need at least two hosts and shared storage (or replicated local storage). When one host fails, VMs automatically start on the second one — typically within 1–3 minutes. For critical applications (e.g., ERP, telephony), this is the difference between one outage per year and 99.9% availability.

Backup and Disaster Recovery

The golden 3-2-1 rule applies here too: three copies of data, on two different media, one copy off-site. In practice, we use Veeam Backup & Replication or Nakivo, with copying to Azure Blob storage (immutable) for ransomware protection. A regular test restore is mandatory — at least quarterly.

When to Consider Moving to the Cloud

If hardware is reaching end-of-life and energy costs are rising, moving to Azure or a similar platform often makes economic sense. Read our article Azure for small businesses, where we compare scenarios in detail.

How many VMs can fit on one physical server?

It depends on the workload. For office applications, we can host 40–60 VMs on a modern server with 32 cores and 256 GB RAM; for SQL or VDI, significantly fewer (4–15).

Is Proxmox production-ready?

Absolutely. Proxmox VE powers production environments for many Slovenian companies. We recommend a paid subscription for access to the stable repository.

What about old applications that need Windows Server 2012?

Virtualization is a great solution — isolate the older OS in a VM that has no direct internet access and stricter firewall rules. However, look for a replacement in the long term.

Do VMware licenses still suit small businesses?

Following the acquisition by Broadcom, the new subscription schemes are significantly more expensive for smaller customers. Therefore, we often recommend Hyper-V or Proxmox for new projects.