So I was thinking about how you would go about implementing an entirely Linux system in to an organisation like a school or government office where they already heavily rely on Windows software. Lets say the school has some specialist software that will won't run under Wine.

The solution I came up with was a shared environment.

Lets say all the legacy Windows terminals are replaced with Ubuntu or perhaps some Thin clients that run from a virtual machine server. It will have as much of the software that will allow it to run as a replacement version of Windows, so Open Office, e-mail client and as much of the Windows only software running in Wine. But it will also have a shortcut to a virtualised Windows service.

This shortcut opens up a remote desktop session to a Windows machine in the virtual machine server running a finite amount of Windows guests. It would make the user think that the client machine they have just loaded up is a normal Windows session, you just have to go through Linux to do it.

Encourage users to use Linux

You will of course, have to encourage your users not to use Windows. So before you load up this Windows session they will be given a friendly box that suggests the Linux way of doing things. It could perhaps suggest that rather than Microsoft Office they should use Open Office, and direct them in how to use it. Rather than IE they should use Firefox. Both of which are cheaper for the organisation and more easily available to the user.

The other thing getting the users to use Linux is the limited amount of Windows machines and licences. When all the Windows machines are used up, you have to use the Linux interface. This should encourage the users to avoid using the Windows environment in case they hog up the finite Windows resource, or for fear of not being able to use a Windows machine and having to suddenly learn the Linux client at a crucial time. There should also be lessons in using Linux and how to make the switch, and why.

Solution Design

During the solution design phase as many of the Windows applications should be tested to see if they work under Wine.

The amount of Windows licenses could also be calculated to a fraction of the required licenses for an entirely Windows environment (say 30%), this will keep costs down and make up for the extra hardware needed, since the Linux cost is close to free.

Realisation

Essentially you turn XP in to a shared elastic resource like most network licensed software is these days. It's when you think of Windows in this way that you suddenly realise how costly it is to have hundreds of machines not being used for 2/3rds of the day.

Of course you now have to manage two environments, but if the XP machines are reverted to the same snapshot each time the user logs off you only need to consider the addition of new apps as a service. Of course you should shy away from implementing new Windows apps so that there is no requirement for a deployment system. Open Source alternatives should be sought at all times.

We all accept that a lot of people won't change to Linux immediately, and there is a lot of data locked up in proprietary formats. This system gives you the ability to bring Linux to an organisations - keeping costs down - and still keep the legacy Windows component. It also allows users to have their own personal phased switch over, going at the speed they want.

Eventually everyone will be working on Linux.