Virtualisation

Jonni Bidwell learns to master containers and VirtualBox, enabling him to run Linux any time, any place and anywhere…

- Jonni Bidwell

Virtualisation has been around since the mainframes of the 60s, where the activities of one program were separated from another. Later, IBM’s CP-40 introduced the notion of a hypervisor and the ability to run multiple OSes concurrently.

Virtualisation proper took off in earnest in the mid-2000s, when 64-bit processors appeared with explicit features for running guest OSes more efficiently. Being able to virtualise machines (in theory) made sysadmins’ lives much easier. Whole systems could be snapshotted, backed up and restored as easily as files. Critical updates could be tested in a virtual sandbox, which vastly reduced the possibility of things catching fire when they were rolled out to physical systems. Multiple VMs could exist on the same system, yet for all intents and purposes be isolated from one another, improving security and efficiency. Home users as well could enjoy the benefits of trying out this “Lye-nux” thing without risking ruination of their incumbent OS.

The hardware has evolved even more since, and you can now pass whole devices to virtual machines (VMs). This makes possible, among other things, running a Windows VM with its own fully accelerated, dedicated graphics card.

After VMs came containers, which rather than implementing a whole OS re-use the host’s kernel and contain just the bits needed to run a particular service or set of services. This enables them to ship with the libraries they require, obviating the problem of conflicting versions when software is installed on a different machine. This makes them more portable than VMs, and to some extent offers a similar level of isolation.