Open Source or Proprietary: Four Public Migrations and the Cost Studies That Do Not Exist

How CodoraTech is funded: CodoraTech is supported by advertising and, in some articles, by affiliate links. Where an article contains affiliate links, we say so at the top of that article.

Munich’s LiMux project put more than 15,000 of roughly 18,000 city desktops onto Linux by October 2013, and in November 2017 the city council voted to return to Windows 10. The French Gendarmerie Nationale, running a much larger estate, reached 103,164 Ubuntu workstations by June 2024 and upgraded them again that December. Both statements are true, which is why the choice between open-source and proprietary software is not settled by any single migration story. The thing most often quoted in this argument is also the thing that is missing: a credible independent comparison of total cost.

Key facts

• Munich: more than 15,000 of about 18,000 city desktops on LiMux by October 2013
• The reversal cost roughly 100 million euros, about half licences and half template and macro conversion, The Register reported
• Schleswig-Holstein moved more than 44,000 mailboxes and 110 million emails and calendar entries
• Gendarmerie Nationale: 103,164 workstations, about 97 percent of its estate, by June 2024
• No independent peer-reviewed total-cost study was found for either side of the comparison

Munich migrated, reverted, and then turned partly back

Munich commissioned a comparison of Windows XP and Linux in 2003. LiMux formally launched in 2004, paused over software-patent concerns and resumed in late 2006. By May 2009, 1,800 workstations were running LiMux and 12,000 had OpenOffice; by October 2013 more than 15,000 of roughly 18,000 city desktops had migrated. City officials claimed savings of 11.7 million euros, about 16 million US dollars, with a later figure of around 10 million euros also cited. That is the city’s own claim, and it was publicly disputed at the time by analysis commissioned by Microsoft.

In November 2017 the council voted to revert to Windows 10 by 2020, citing user dissatisfaction and limited availability of Linux-compatible specialist software. The Register reported the cost of that reversal at approximately 100 million euros in total, around 50 million for Windows 10 licences and around 50 million to convert LibreOffice templates and macros and build a new Microsoft Office templating system. The honest nuance is in the review that Accenture authored, which found that organisational issues rather than technical limitations of Linux were the most important causes of the problems. The reversal is not clean evidence that open source cost more.

Nor was it the end. In May 2020 a new Green and SPD coalition shifted the city back toward open source where technologically and financially possible, and in October 2024 Munich allocated 200,000 euros for an Open Source Program Office. Anyone citing Munich as a closed case, in either direction, is citing a round trip.

Forty-four thousand mailboxes, and a motive that is not cost

Schleswig-Holstein is the live example. According to the European Commission’s Open Source Observatory, it is the first German federal state with a formal Open Innovation and Open Source Strategy, published in November 2024, with the migration effort itself beginning in 2020. It has completed the move of the state administration email system: more than 44,000 mailboxes and 110 million emails and calendar entries transferred to Open-Xchange and Thunderbird. Approximately 80 percent of the office ecosystem has moved to LibreOffice, with a plan to put all work laptops on Linux and adopt open-source cloud storage.

The state claims 15 million euros saved in licence fees, of which 9 million euros was reinvested into the open-source ecosystem after switching costs. That figure originates with the state and is relayed by an observatory funded to promote open source, so it should be read as an interested party’s number rather than an audited one. More useful is the stated motive, which is not cost at all: digital sovereignty, transparency, informational symmetry and privacy. Note also what the write-up does not contain. It discloses no technical or organisational difficulties encountered, and in an advocacy-adjacent source the absence of reported problems is not evidence that there were none.

The Gendarmerie’s estate reached 103,164 workstations

The strongest counterweight to Munich is larger than Munich, and it stuck. The French Gendarmerie Nationale has run its own Ubuntu derivative, GendBuntu, for well over a decade, and the deployment curve is public.

1. 2004: OpenOffice replaces Microsoft Office on 20,000 computers
2. 2008: decision taken to migrate to Ubuntu
3. January 2011: large-scale GendBuntu 10.04 LTS rollout begins
4. November 2011: 35,000 workstations; December 2013: 43,000; June 2014: 65,000
5. March 2017: 70,000 workstations; June 2024: 103,164, about 97 percent of the estate
6. December 2024: upgraded to GendBuntu 24.04

The cost comparison nobody has produced

The Gendarmerie has claimed a 40 percent reduction in total cost of ownership, stated in December 2013, and estimated software-cost savings of roughly 2 million euros a year. That is the Gendarmerie’s own claim, published through the European Union’s Open Source Observatory and Repository. Every headline saving in this article belongs to the organisation that made the switch, and no independent verification of any of them was found.

This is not an accident of research. The Register’s assessment, written while Munich was reversing, is that figuring total cost of ownership between proprietary and open-source software is a guessing game, that reliable Windows-versus-Linux studies are lacking, and it quotes Italo Vignoli of the Document Foundation saying that most research is sponsored by proprietary software companies and as such might be biased. No independent, non-vendor, peer-reviewed study comparing the two at organisational scale was found for this article, and a documented migration in the other direction with published cost figures does not appear to exist outside Munich.

That absence changes what a buyer should ask. Since the cost answer is unavailable, the useful questions are the ones the documented cases actually turned on.

• Does specialist software your department depends on run on the target platform, which is the reason Munich’s council gave
• How large is the template and macro estate, given that converting Munich’s cost roughly as much as the licences
• Who is funding any comparison study you are shown, and does it disclose that
• Is the goal cost or sovereignty, as Schleswig-Holstein states, because the two justify different decisions
• Is the organisation prepared to run the change as an organisational programme, which is where the Accenture review located Munich’s problems

Sources: Wikipedia · The Register · European Commission Interoperable Europe · Wikipedia · Open Source Observatory and Repository