Virtualisation is one of the most transformative technologies in the history of enterprise IT. By abstracting hardware resources into software-defined layers, virtualisation allows organisations to run more workloads on less hardware, provision new environments in minutes, and build resilient infrastructure that would require physical data centres to replicate. This guide covers the full virtualisation landscape and how to maximise its value.
What Is Virtualisation?
Virtualisation creates a software abstraction layer between physical hardware and the operating systems or applications running on it. A hypervisor (the virtualisation engine) allows multiple virtual machines (VMs) to share the resources of a single physical server, each operating as if it has dedicated hardware.
Types of Virtualisation
| Type | What Gets Virtualised | Key Technology |
|---|---|---|
| Server virtualisation | Physical servers → multiple VMs | VMware vSphere, Hyper-V, KVM |
| Desktop virtualisation | End-user desktops → centralised VDI | Citrix DaaS, VMware Horizon, AVD |
| Network virtualisation | Physical network → virtual overlays | NSX-T, ACI, OpenStack Neutron |
| Storage virtualisation | Physical storage → software-defined pools | VMware vSAN, NetApp ONTAP, Ceph |
| Application virtualisation | Apps delivered without local install | Citrix Virtual Apps, App-V |
| Container virtualisation | OS-level process isolation | Docker, Kubernetes, containerd |
Server Virtualisation: The Foundation
How It Works
A Type 1 hypervisor (bare-metal) installs directly on physical hardware and manages multiple VMs. Each VM gets a share of CPU, RAM, and storage allocated by the hypervisor.
- VMware ESXi: Industry-leading enterprise hypervisor, managed through vCenter
- Microsoft Hyper-V: Built into Windows Server, managed via System Center or Windows Admin Center
- KVM (Kernel-based Virtual Machine): Open-source, Linux-based, foundation for OpenStack and many cloud providers
Key Benefits
Hardware consolidation: A modern server can typically support 20–40 VMs. Replacing 30 physical servers with 3 virtualised hosts reduces:
- Power consumption by 70–80%
- Cooling costs proportionally
- Physical data centre space requirements
- Hardware maintenance overhead
High Availability and Fault Tolerance:
- Live migration (vMotion/Live Migration): Move running VMs between hosts with zero downtime for patching, load balancing, or hardware maintenance
- High Availability (HA): Automatically restart VMs on surviving hosts if a physical server fails (typically < 60 seconds)
- Fault Tolerance: Mirror VM execution across two hosts simultaneously for zero-downtime failover
Rapid Provisioning: Deploy a new server in minutes from a template rather than waiting weeks for hardware procurement and physical installation.
Snapshot and Backup: Take point-in-time snapshots of VMs before risky changes, enabling instant rollback. Simplifies backup with VM-level consistent snapshots.
Resource Management Best Practices
Capacity Planning
- Target average CPU utilisation of 60–70% across the cluster (headroom for bursts and HA failover)
- Avoid over-committing RAM by more than 1.2:1 (memory is less compressible than CPU)
- Monitor storage I/O latency; high latency (> 20 ms) indicates storage bottlenecks
Right-Sizing VMs
Over-provisioned VMs waste host resources that could be used by other workloads. Review and right-size VMs regularly:
- VMs using < 20% of allocated CPU for extended periods should be reduced
- Use VMware vRealize Operations or similar to identify right-sizing opportunities
- Implement memory ballooning and transparent page sharing where supported
VM Sprawl Prevention
Uncontrolled VM sprawl is a common problem—it is easy to create VMs and forget about them:
- Implement a VM lifecycle policy (creation requires approval; unused VMs auto-power-off after 30 days)
- Regular audits to identify powered-off VMs older than 90 days
- Integrate VM provisioning with CMDB to maintain an accurate asset inventory
Desktop Virtualisation (VDI)
Virtual Desktop Infrastructure (VDI) centralises desktop computing in the data centre, delivering virtual desktops to end users over the network:
- Persistent VDI: Each user has a dedicated, persistent VM (higher storage cost, better user experience)
- Non-persistent VDI: Users log into a shared pool VM that resets at logout (lower cost, limited personalisation)
- Session-based (RDSH): Multiple users share a single OS instance (most cost-efficient, limited isolation)
When VDI Makes Sense
- Highly regulated environments requiring data to never leave the data centre
- Knowledge workers needing access from any device/location
- High-security workloads (financial traders, healthcare)
- Simplified endpoint management at scale
Azure Virtual Desktop (AVD)
Cloud-based VDI eliminates the need to manage VDI infrastructure. AVD delivers Windows 11 desktops from Azure with:
- Pay-per-use pricing (no capital investment)
- Global availability (low-latency access for distributed teams)
- Integration with Microsoft 365 and Microsoft Intune
- Significant cost savings for organisations already in the Microsoft ecosystem
Network Virtualisation
Software-Defined Networking (SDN) and Network Function Virtualisation (NFV) virtualise network infrastructure:
- Overlay networks: Create logical network segments (VXLANs) independent of physical topology
- Virtual firewalls and load balancers: Deploy as software appliances rather than dedicated hardware
- Micro-segmentation: Apply granular security policies between VMs without physical network changes
VMware NSX-T is the market leader for enterprise network virtualisation, enabling security policies that follow workloads regardless of where they run.
Storage Virtualisation
VMware vSAN and similar hyper-converged infrastructure (HCI) solutions virtualise local server storage into shared storage pools:
- Eliminate dedicated SAN hardware
- Linear scalability (add nodes to increase both compute and storage)
- Built-in deduplication and compression reduce effective storage costs
Measuring Virtualisation Success
Track these KPIs to demonstrate value:
| KPI | Target |
|---|---|
| Server consolidation ratio | 15:1 or higher |
| Average host CPU utilisation | 60–70% |
| VM provisioning time | < 15 minutes from template |
| HA recovery time | < 60 seconds |
| Power consumption reduction | 60–80% vs physical baseline |
| Hardware refresh cost reduction | 50%+ per compute cycle |
Virtualisation is no longer an advanced capability—it is table stakes for any organisation running more than a handful of servers. The organisations that maximise its value are those that manage it proactively, avoid sprawl, and extend its principles into cloud and container environments as their infrastructure evolves.