Several states of India are moving their governments rapidly to FLOSS in response to XP and just because it’s the right way to do IT. Kerala has recently published a manual. Here’s an excerpt:
“1. Equip internal IT teams to plan the migration (in case there is no internal IT team, identify a service provider or technology partner who would provide assistance for the entire process).
2. Inventory the installation (number and configuration of PCs, devices and applications) and create an initial plan for migration.
3. Back up all user data (including application data, email, documents etc) on existing machines to back-end, offline media or the Cloud.
4. Choose an appropriate GNU/Linux Distribution keeping in mind the requirements of the organization.
5. Enumerate the legacy hardware (old desktops, printers, scanners etc) and determine which have compatibility issues with the chosen distribution.
6. Enumerate the legacy software applications and identify which of them cannot be used on the GNU/Linux distribution directly or indirectly. This may include local applications (eg., browsers, office suites) and networked applications (file/print sharing, browser-based applications, client-server applications and networked applications).
7. Map the Windows desktop applications used in the system to available FOSS equivalents (See Section , Item on Page 12). Consider piloting alternatives before finalizing.
8. Categorize the installation into three groups: (a) the desktops that can be immediately migrated; (b) those that require low-effort porting; and (c) those that cannot be moved immediately due to various reasons.
9. Plan and implement the installation of the chosen distribution and the capacity building and training programmes for users as well as IT staff in a synchronized fashion. In particular, equip users to handle immediate and routine tasks such as browsing, printing, booting, and handling documents. Another priority area is to equip users to know the differences between FS office applications (LibreOffice, OpenOffice) and proprietary applications (such as MS Office) for routine tasks.
10. Set up a Helpdesk (internal or external) for hand-holding new users.
11. Review progress periodically and if there are significant issues, seek external resolution.”
They call it A HANDBOOK ON MIGRATION OF DESKTOPS TO FREE SOFTWARE PLATFORMS. I call it the epitaph of Wintel in government. There’s no hand-wringing about what to do: “A basic question that is often posed is that while migration to Free Software platforms for the individual or small groups is possible and easy, whether the same can be said of large institutions, particularly the Government. The answer to this question is that such migration can indeed be done, provided a structured and systematic approach is followed.” They just get on with it. Wintel just isn’t necessary. A few items seem necessary only because of lock-in.
A Quick Note on the Title Wikipedia: “Crushers may be used to reduce the size, or change the form, of waste materials so they can be more easily disposed of or recycled”
UPDATE I sent feedback to ICFOSS about their handbook, suggesting they look into thin-client technology to speed up the process of migration, increase performance and to reduce cost/performance. Here’s the text:
“The handbook is pretty good as it stands but I suggest a couple of changes:
“Any large-scale migration of this kind carries with it risks which need to be carefully evaluated and mitigated. Migration to a FS platform is not a “fit-and-forget” process, but requires evaluation of the desktop installation in each enterprise, identification of problematic areas (legacy hardware, desktop applications, training & capacity development requirements, to name a few), developing a migration plan, and determining mitigation measures” is much too harsh. Just migrating to the next version of Windows has exactly the same risks unless you buy the OS preinstalled which hides its price. This paragraph should be balanced with an outline of the risks of not migrating: further lock-in, slowing down, malware, re-re-reboots and breaking of open standards. I believe the longer one uses that other OS, the harder it is to escape. Compare Munich’s experience which took a decade to escape compared to Extremadura which installed GNU/Linux over a weekend for all of its education and many other positions.
Further, there is no mention of thin-client technology. Modern PCs and servers are so powerful they can run all the processes for 30-1000 users simultaneously with GNU/Linux. The X Window System makes it easy to install software on some powerful/newer/robust/resourceful machine and users can interact from a small cheap thin client over a network. This greatly reduces the cost of a new installation and can be used to migrate very quickly. The software only needs to be installed on a few machines and the old machines can be set to boot over the network eliminating the need for the hard drive to be powered (PXE). See LTSP.org. There are fancier ways to do this but this is the most efficient. For instance, the server can cache most of the files needed to run the applications, login or open a window in RAM eliminating all those seeks on hard drives. Typically this speeds up the performance by a factor of 3 to 5 and users get that speedup on the old hardware… Seeing machines go faster is a great motivator and overcomes a lot of personal resistance to change.
Another similar but more voluminous handbook is IBM’s Migration Cookbook. It, too, is very conservative but they describe the typical scenarios in greater detail and suggest 80% of migrated PCs are in the “easy” category because many users use just a few applications like browser/office suite.
If you want to investigate thin clients in greater detail, check out what Largo, Florida does. They have a few specialized servers running hundreds of users simultaneously. e.g. They will have one server just doing logins. Another will run FireFox and another runs LibreOffice. The user sees windows generated by those servers on their local screens and has no idea the work is spread out over many servers thanks to network forwarding through SSH and similar technologies. A few servers working hard cost less than $100 per user and a cheap thin client may cost only $100 so the total cost of the system per user is $200 and, thanks to file-caching on the servers, the performance exceeds any thick-client PC.
If you are going to the trouble of changing the OS and applications, you should also consider changing the whole architecture and management. That will save most of the work of migrating and speed it up. That’s how you can switch over on a weekend: set up the servers and test them as long as you want, weeks or months, then set up all the machines to boot PXE from the servers on their LANS, issue account userids and passwords and you’re done. The last such migration I did switched 100 PCs and 500 users in one month from start to finish. The project was on time, under budget and worked perfectly. With FLOSS, you can generate userids automatically, hundreds per hour, even on modest hardware.
Have server, will travel… “