Server virtualisation is the use of software and hardware-abstraction techniques to run multiple logically independent server environments on a single physical server or across a pool of physical resources. Instead of assigning one physical machine to one operating-system instance, virtualisation allows a physical host to provide virtual CPUs, memory, storage and network interfaces to multiple virtual machines. Each virtual machine can run its own operating system and applications while remaining logically isolated from other virtual machines. A hypervisor manages this abstraction and allocates physical resources to the virtual machines. Server virtualisation can improve hardware utilisation, simplify provisioning, support isolation and make workloads easier to move, replicate or recover. It is a foundational technology in modern data centres and cloud computing.
Before widespread virtualisation, servers were often deployed directly on physical hardware, meaning that a workload could occupy an entire machine even when it used only a fraction of the available CPU and memory. Different applications could also require incompatible operating systems or software environments, encouraging organisations to maintain separate physical servers. Virtualisation allows these workloads to share a host while remaining separated at the operating-system level.
The development of hardware-assisted virtualisation features by CPU manufacturers, improvements in memory management and advances in storage and networking made virtual machines increasingly practical at scale. Data centres consequently moved from predominantly one-server-per-workload models towards consolidated virtual infrastructure and later towards combinations of virtual machines, containers and managed cloud services.
The hypervisor is the central software layer in server virtualisation. A type 1, or bare-metal, hypervisor runs directly on the physical host hardware and manages virtual machines without requiring a conventional general-purpose host operating system underneath it. A type 2 hypervisor runs as an application on a host operating system and is common in desktop or development environments. Enterprise server virtualisation typically uses type 1 architectures because they provide direct control over hardware resources and are designed for high-density, managed workloads. The hypervisor presents virtual hardware to each guest operating system and mediates access to physical processors, memory, storage and networking.
CPU virtualisation allows several virtual machines to share physical processor resources. The hypervisor schedules virtual CPUs onto physical CPU cores or threads and manages privileged operations that guests cannot safely perform directly. Modern processors provide hardware virtualisation extensions that make this more efficient by supporting guest execution and controlled transitions between guest and hypervisor contexts. Memory virtualisation is similarly important.
The hypervisor maintains mappings between guest physical memory and host physical memory and may use mechanisms such as nested page tables to accelerate translation. Memory overcommit can allow administrators to allocate more virtual memory than the host physically contains, but this depends on techniques such as ballooning or swapping and can degrade performance when actual demand exceeds available physical memory.
Storage and networking are also virtualised. A virtual machine can see a virtual disk that is backed by a file, logical volume, network storage system or another storage abstraction. Features such as snapshots, thin provisioning, cloning and virtual disk migration can simplify management, although snapshots are not substitutes for independent backups and can create performance or storage-management issues when retained improperly. Virtual network interfaces connect VMs to virtual switches and then to physical network interfaces or other virtual networks. Virtualisation can therefore reproduce many features of physical networking, including VLANs, routing, firewalls and software-defined network policies, while keeping the logical infrastructure independent of individual physical devices.
One major benefit of server virtualisation is workload consolidation. Multiple moderately utilised servers can run on a smaller number of physical hosts, reducing hardware, rack space, power and cooling requirements. Virtual machines can also be provisioned from templates, cloned for testing and migrated between compatible hosts. Live migration technologies can move a running workload between physical hosts with limited interruption, which can assist maintenance and load balancing. High-availability systems can automatically restart virtual machines on another host after certain host failures. These capabilities improve operational flexibility, although they introduce dependencies on shared storage, network infrastructure, cluster management and compatible hardware or hypervisor configurations.
Virtualisation has trade-offs. The hypervisor and virtual hardware introduce some overhead, although modern virtualisation can approach native performance for many workloads. Oversubscription can cause contention when too many VMs compete for CPU, memory, storage or network resources. A physical host failure can affect many virtual machines simultaneously, creating a larger failure domain than isolated physical servers. Security also requires protecting the hypervisor and management plane because compromise at that level can potentially affect multiple guests. Licensing, operational complexity, monitoring and capacity planning are additional considerations. Virtualisation should therefore be viewed as resource abstraction and isolation technology, not as an automatic performance or reliability improvement independent of architecture.
Server virtualisation differs from containerisation because a virtual machine normally virtualises hardware sufficiently to run a complete guest operating system, while containers generally share the host operating system kernel. It also differs from cloud computing: cloud services can use virtualisation internally, but cloud computing additionally encompasses service delivery, automation, APIs, elastic provisioning and managed infrastructure. Server virtualisation remains useful in on-premises data centres, private clouds, public clouds, development environments and disaster-recovery systems. Its central architectural contribution is the decoupling of server workloads from specific physical machines, allowing compute, memory, storage and networking resources to be allocated dynamically to multiple isolated operating-system environments.
