How Does Server Virtualization Work?

Virtualization at a Glance

In corporate IT environments, servers are needed to do many jobs. Each machine is tasked to do only a single job, and often many servers are given the same job. This is done to prevent hardware and software problems on one machine from endangering several systems. There is a downside with this arrangement, though, in that server hardware can be very expensive, and many servers do not utilize their full resources. When this is the case, server virtualization can provide a way to reduce hardware costs while still providing protection against software failure.

Virtual Servers

The fundamental unit of software virtualization is a virtual server. This is essentially a complete running PC set up to perform a server function. It is different from a PC, though, in that it is run from within another operation system. In this way, dozens of virtual servers can be run on a single set of server hardware. Each virtual server is independent of the hardware it is run on, so that it can be run on any physical machine without modification. Virtual servers must share the resources of a server, so attention must be paid to what level of resources each virtual server provides. If a virtual server encounters a software error that would normally cause the server to shut down, it can shut itself down without affecting other virtual machines on the physical server.

Machine Supervisors

When a virtual server encounters an error and is forced to shut down, it needs to be restarted and brought back into operation. To accomplish this, and to manage server resources, a program called a machine supervisor or sometimes a hypervisor is used. This software allocates resources to each virtual server and makes sure that it is in operation, restarting servers as necessary.

System Redundancy

While server virtualization can continue to provide protection against software-based failures, it actually increases the danger of hardware failures. If a server cannot remain functional as a result of hardware failure, all virtual servers on that machine are impacted. To offset this problem, physical servers are often run in clusters using a single machine supervisor. If one machine is overburdened or experiences a hardware failure, the machine supervisor will automatically restart those virtual servers on a new physical host to prevent interruption. This provides a high degree of reliability and flexibility for virtual server setups.