A hypervisor is a form of technology that enables virtualised services. Also known as virtual machine managers/monitors (VMMs), hypervisors provide computer systems with a specialised layer of software that allows hardware to be utilised by multiple virtual machines (VMs).
Hypervisors and virtualisation: a shared history
The history of hypervisors is bound up with the concept of virtualisation. The technology first emerged in the 1960s and 70s, when ‘time-sharing’ was developed as an innovative way of multitasking and sharing resources between several users on mainframe computers.
It was around this time that the term ‘hypervisor’ originated as a stronger version of ‘supervisor’ – which is itself an older term for an operating system kernel. In this sense, the original role of a hypervisor was to supervise the supervisor.
A key element of virtualisation
Fast-forward to the mid-2000s, when advances in hardware and software allowed computer resources to be virtualised relatively easily. Hypervisors now performed the vital task of abstracting physical hardware and separating it from operating systems and applications.
Today, hypervisors see widespread use running VMs on shared server hardware. A hypervisor manages the allocation of physical resources, supervises each 'guest' operating system’s access to the underlying hardware, and allows virtual machines to function as separate, isolated systems.
Types of hypervisor
Hypervisors can be divided into two main categories. Type 1 hypervisors, such as VMware ESXi/vSphere and Microsoft’s Hyper-V, are often called 'bare-metal' or 'native' hypervisors because they run directly on top of the host hardware. This is as opposed to type 2 or 'hosted' hypervisors like VMware Workstation and VirtualBox, which are installed onto an existing OS.
With no additional OS overhead, type 1 hypervisors are usually faster and more secure, making them more suited to data centres and large-scale cloud applications. On the other hand, type 2 hypervisors are generally easier to install and manage, with the benefit of an OS already in place.
Hypervisors and containers
The virtualisation provided by a hypervisor differs from operating-system-level virtualisation, or containerisation. Containers share a single OS kernel, so a complete guest OS isn’t needed for each containerised application.
This means that, a lot of the time, it makes sense to run an application in a container, not a full-blown virtual machine. But hypervisors can still have an important role to play, with containers themselves often running on VMs.
So it’s more a case of containers and hypervisors finding synergy together, rather than one replacing the other. For example, instead of running each application on its own virtual machines, a more efficient setup could utilise fewer VMs, each hosting multiple containerised applications.
To virtualise or not to virtualise
If you need a server hosting solution, choosing between a virtualised package and physical hardware is a major decision. Virtual private servers and cloud servers rely on hypervisors to maximise the flexibility of hosting resources. Not only does virtualisation allow servers to be dynamically scaled, it also provides an economy of scale that makes cloud hosting more affordable.
But hypervisors aren’t without downsides. The hypervisor software layer adds another overhead on the server, soaking up extra processing power and memory that would otherwise be spent on your web projects. This means that a virtual server can’t offer the same performance as a dedicated or bare-metal machine with equivalent resources.
Without the extra overhead of a hypervisor, dedicated hardware can deliver undiluted, raw power for running websites and applications. But this comes at a significantly higher cost, with the added limitation that physical servers are less scalable and customisable than virtual machines, which can be deployed or reconfigured in a matter of seconds, not hours.
Should you use a virtualised setup?
Overall, a virtualised infrastructure is excellent when flexibility is critical – e.g. if resources need to be scaled up and down to match fluctuating demand and traffic spikes – and when cost control is a top priority. Physical hardware is better suited to long-term, consistently high workloads, where the hypervisor layer could add an unnecessary drain on performance.
Of course, there’s nothing to stop you from mixing and matching virtual machines with dedicated boxes to create the ideal combo of flexibility and performance. Our CloudNX platform provides dedicated Bare Metal machines – with the ability to run your physical machines alongside load-balanced VMs in a unified infrastructure. For more details on how to create your ideal hosting setup, get in touch with our experts.