Cloud Architecture

Types of Virtualization in Cloud Computing With Examples

The main types of virtualization in cloud computing, with examples of server, storage, network, desktop, and container-based virtualization.

On this page
  1. What virtualization means in cloud computing
  2. Server (hardware) virtualization
  3. Storage virtualization
  4. Network virtualization
  5. Desktop virtualization
  6. Application virtualization
  7. Operating-system virtualization (containers)
  8. Data virtualization
  9. What is the impact of virtualization on cloud computing?
  10. Choosing a type

What virtualization means in cloud computing

Virtualization creates a software version of a physical resource, such as a server, storage device, network, or desktop. One physical resource can be divided into several virtual ones, or several can be pooled into one. Most cloud infrastructure depends on it. For where virtualization sits in the software stack, see the levels of virtualization.

Virtualization and cloud computing are not the same thing. Cloud adds self-service, elastic scaling, and metered billing on top of virtualized resources.

Server (hardware) virtualization

A hypervisor runs several virtual machines (VMs) on one physical host. Each VM has its own guest operating system and applications. This is the type most people mean when they talk about virtualization in cloud computing.

Type 1 and Type 2 hypervisors

  • Type 1 (bare-metal) runs directly on the physical hardware. Examples include VMware ESXi, Microsoft Hyper-V, Xen, and KVM.
  • Type 2 (hosted) runs on top of an existing operating system. Examples include Oracle VirtualBox, VMware Workstation and Fusion, and Parallels Desktop.

A Type 1 hypervisor does not mean there is no operating system. KVM is part of the Linux kernel, and Hyper-V uses a Windows parent partition. The Type 1 and Type 2 split describes the design. It is not a fixed ranking of speed for every workload.

Techniques

  • Full virtualization runs an unmodified guest operating system. Early x86 processors made this difficult, so software techniques such as binary translation, which VMware used from 1999, were needed.
  • Paravirtualization changes the guest operating system so it calls the hypervisor directly. Xen, introduced in 2003, is a well-known example.
  • Hardware-assisted virtualization uses CPU features such as Intel VT-x and AMD-V, which appeared around 2005 and 2006.

Storage virtualization

Storage virtualization pools physical disks into logical volumes. Users see one pool rather than individual disks, so capacity can grow or move without changes to the applications using it. Examples include storage area networks (SANs) and software-defined storage such as VMware vSAN. Cloud block storage services use the same basic idea.

Network virtualization

Network virtualization creates and manages network resources in software. Common building blocks include:

  • VLANs (IEEE 802.1Q), which split one physical network into separate logical networks. The format allows 4,094 usable VLAN IDs.
  • VXLAN, which uses a 24-bit segment ID and allows about 16 million segments.
  • Software-defined networking (SDN), which separates network control from packet forwarding.
  • Network functions virtualization (NFV), which runs functions such as firewalls and load balancers as software.

VMware NSX is one example of a network virtualization product.

Desktop virtualization

Desktop virtualization, often called virtual desktop infrastructure (VDI), runs desktop environments on servers and delivers them to user devices. Examples include Citrix, Amazon WorkSpaces, and Azure Virtual Desktop. Users can work from different devices, and IT staff can manage software and updates in one place.

Application virtualization

Application virtualization runs an application in an isolated environment, or streams it from a server to the user’s device. Examples include Microsoft App-V and Citrix Virtual Apps.

Operating-system virtualization (containers)

Operating-system virtualization runs several isolated user spaces on one kernel. Docker and LXC are examples. Containers share the host’s kernel instead of running their own, which makes them lighter than VMs. The containers versus VMs comparison covers the differences in isolation and startup time.

Data virtualization

Data virtualization gives one view of several data sources without copying the data into a new store. Denodo is one example of a data virtualization product.

What is the impact of virtualization on cloud computing?

Virtualization changed how cloud providers run their infrastructure in five main ways:

  • Multi-tenancy. Many customers can share the same physical hardware, each in an isolated virtual environment.
  • Higher utilization. Before virtualization, many physical servers often ran at low utilization. Estimates vary, but figures of roughly 5 to 15 percent are often cited. Consolidating workloads onto fewer hosts uses hardware more efficiently.
  • Faster provisioning. A new VM can be created in minutes, rather than waiting for hardware to be ordered and installed.
  • Live migration. A running VM can move between physical hosts, for example with VMware vMotion. This helps with maintenance and load balancing.
  • Snapshots and cloning. Administrators can capture a VM’s state and copy it for backup, recovery, or testing.

Virtualization also has trade-offs. The hypervisor adds some overhead. Workloads on the same host can compete for resources, an effect sometimes called the noisy neighbor problem. Security depends on the isolation the hypervisor enforces, so hypervisor flaws and weak management access are serious risks. Some vendors also license software per virtual CPU or per core, which can raise costs as an environment grows.

Choosing a type

Most cloud customers never install a hypervisor. They choose instances, storage, and networks, and the provider handles virtualization underneath. Knowing the types still helps when you read a provider’s documentation. Providers group VMs into families and types, and a virtual CPU is a slice of a physical processor, not a whole core.

The complete guide to cloud architecture

Cloud Computing Architecture: The Core Components
Read the guide

Keep exploring

Guides from other topics

All topics →