Over the years I have read many accounts of migrations large and small. They big guys tend to be very cautious because any disruption is huge. This is what Canonical says about how to go about it:

1. Plan effectively for maximum effect
2. Target the users ripe for migration
3. Identify the apps that save you money and fuss
4. Create the right management flows
5. Pilot your project to get it just right

That’s not too different from what

• IBM says, and
• Munich says

see FIVE GOLDEN RULES FOR A SUCCESSFUL UBUNTU DESKTOP MIGRATION

While a perfectly planned set-piece migration appears to work for large organizations, smaller organizations may simply experience delay and greater costs doing the detailed work. The GNU/Linux desktop has evolved to the point where for a large proportion of users it can do the job with little fuss. Just backup data, install the OS and restore the data. If any problems arise they are likely to be small and manageable. With a good backup, one can always revert particular machines if a show-stopper arises. In ten years of migrating small organizations I never encountered a show-stopper that could not be simply worked around. Migrations of simple computer labs may take only an hour or two. A whole school may be about as complicated as that. Where I last worked, I walked around replacing PCs with GNU/Linux PCs. I could have installed over the network to avoid the walking but there was a matter of locked doors after hours… That’s not a show-stopper associated with the OS, just constraints on the institution.

For the benefits of global mirrors, a great package manager, APT, good tools for system management over the network and sound policies on integration, I recommend Debian GNU/Linux. It works for you.

“

An important motive for the switch to open source were negative experiences with a proprietary database and the services provided by the company, remembers Kaell, without going into details. As soon as possible the Cadastre started switching to the alternative, Postgresql. "And we’ve gradually expanded from there." ACT now relies heavily on Linux servers, most of them virtualised.

“
That sounds right to me. Any time a government is using taxpayers’ money to buy licences when they could use FLOSS for $0, taxpayers should demand change. Fiducial responsibility of governments should be driving adoption of FLOSS in government. It’s just silly to pay more for a product or service than the cost of production which is what M$’s OS is all about and Oracle’s database and Adobe’s graphics software too. There are all FLOSS alternatives for these widely used products. Waste on the scale M$ demands whether on client, server or a whole system is colossal. M$ had $11 billion operating profit from revenue of $18 billion for their client OS alone in their last fiscal year. That means the cost of production and distribution was only $7 billion. Of course, FLOSS not being sold would have even lower costs of production. Sharing saves a lot of money. Oracle, the database company had $37 billion in revenue their last fiscal year. Does the world owe such businesses a living? No.

I recommend a distro like Debian GNU/Linux as a basis for IT in government. Debian has been around a long time, is globally developed and distributed, costs $0 to use and is very open about bugs and solutions.

see Luxembourg's cadastre administration prefers to use open source | Joinup.

Look at the specs for Lenovo’s latest:

“LePhone K860 smartphone featuring Samsung Electronics’ quad-core Exynos 4412 processor, 1GB RAM, Android 4.0, 5-inch 1280 by 720 touch screen and 8-megapixel rear camera.”

see Lenovo offers quad-core LePhone K860.

Lenovo sold 5 million smart phones last quarter and plans to sell 20 million in 2012 with 40 models. Tell me they are not having fun… Lenovo sold 12.8 million x86/amd64 PCs last quarter. 5 million smart phones with specs that would embarrass most PCs from a decade ago and still fit in the pocket are a wonderful addition to Lenovo’s product line. Guess which are more profitable after paying M$ a licensing fee for Wintel PCs. Guess which software has the larger margin, FLOSS or that other OS? Guess which give Lenovo real independence from M$’s interference in their business.

Lenovo had huge growth in profit last year and only 16% of revenue comes from North America. They do sell a lot of PCs in regions where GNU/Linux flourishes.

If Lenovo and other OEMs can make a decent living selling FLOSS on smart phones they have a hedge against any disruption in income due to M$’s penchant for disaster like Vista and “8″. They also have a cushion to use if they decide to ship more GNU/Linux PCs. OEMs would be glad to be paid more for their labour making hardware and less for selling M$’s stuff. Lenovo makes PCs of many kinds. They don’t have to make Wintel PCs.