Open Source Total Cost of Ownership 2.0
Published 11:35, 07 December 11
Back in 2006, I wrote a piece for LXer called "A Brief History of Microsoft FUD". This ran through successive attempts by Microsoft to dismiss GNU/Linux in various ways. One of the better-known was a series of "Total Cost of Ownership" (TCO) studies. By an amazing coincidence, these all showed that Microsoft Windows was cheaper than that supposedly cheap GNU/Linux.
Fortunately, people soon cottoned on to the fact that these studies, paid for by Microsoft, were pretty worthless (here, for example, is a great debunking of the kind of FUD that was being put out in 2005.) However, one knock-on consequence of that episode is that TCO studies rather fell from favour.
So it's interesting to see this new report prepared for the Cabinet Office with the title "Total Cost of Ownership of Open Source Software", which has been released under the liberal Open Government Licence for public sector information. Here's the background:
This report is intended to provide a balanced assessment of the potential of open source software within the public sector based on evidence collected from those who have taken this path and from members of the community of firms offering support services to such public bodies. The audience we are addressing includes politicians and senior decision makers across central and local government, senior IT managers and the supplier communities.
And here's how the information was gathered:
This study was structured in two phases. Phase 1 was based around a pro-forma data collection instrument that was completed and returned by 32 people (see Appendix C). The pro-forma was set up for access as an electronic and printable document (available in odt, pdf and doc formats) and an online version in SurveyMonkey. The pro-forma was online for a period of two months. We received twenty-five responses online plus seven returned by email. Phase 2 was concerned with in-depth interviews with 20 people in 14 organisations. In both phases a mix of public sector and private sector organisations were included, with the in-depth studies (phase 2) involving a preponderance of public bodies.
Obviously, a critical issue is what is meant by "TCO" in this context:
Our developed definition is thus, “TCO reflects a measure of all the costs of identifying and acquiring software, away from the software. TCO reflects not just the direct qualities of a software product (price, functionality, reliability) but also the relationship of the software to the organization‟s broader set of technology platforms, installed systems, skills and strategic goals, as well as available market and community based services.”
On that subject, the report makes the following important point:
In many relevant respects open source software is different to proprietary software (both parties would probably agree on that). It places different demands on, and offers different benefits to, the host organisation and it is embedded in somewhat different software ecosystems and is served by different supply chains . The question of TCO thus becomes less focused on what software costs per se to purchase or over its lifetime. TCO has to consider questions of how software fits into the organisation and relates to the other resources including legacy systems, technology platforms and infrastructures, skill sets and management style, as well as business strategy.
This shows how much more nuanced the concept of TCO has become since those Microsoft studies, which tended to focus on more obvious issues.
One of the things that emerges from the report is an appreciation amongst users of free software that the benefits go well beyond simple savings:
early adopters of open source applications in the public sector quote benefits such as reduced vendor lock-in as one of their key arguments alongside lower costs.
That's something that I've always mentioned when people have asked me for the key benefit of using open source - the freedom from lock in and the ability to take control of your computing destiny. It's good to see that coming out here.
I was also pleased to see the following point raised:
Many interviewees explained that open source cost savings materialize mid to long term rather than in the short-term. Thus, they report, it is important to manage expectations to ensure that a project that chooses open source software is not considered a failure prematurely if it does not deliver excellent service at substantially lower cost on day one.
That's crucially important if we are to avoid tales of open source migration "failures": the projects must be given enough time to prove themselves.
When migrating between open source products it is understood that costs may be lower because adherence to open standards allows greater interoperability. The organizations contributing to this study who had a more long term view of their open source software adoption gave comments to the effect that the migration costs (exit costs) were more favourable for open source and so this had for them become one of the deciding factors in favour of OSS.
Those migration/exit costs are almost never considered - certainly not in the early TCO studies pushed by Microsoft - but it's a really important aspect that companies need to bear in mind.
But in many ways the most interesting and exciting result to emerge from this new study is the following:
A facet that is seen as beneficial but rather unexpected is how a culture of innovation and more risk taking behaviour can be promoted as open source is used. Open source adoption has, for example, forced local authorities to become more accepting of "mistakes" that can be identified and rectified quickly by hands-on access to code and configurations. Experience of such agility and empowerment can spur the change in favour of open source.
I've noted before how deadening this fear of failure has been in business and public sector computing, and why uptake of open source has been so slow in this country as a result. The fascinating thing this new study suggests is that once that fear of failure is overcome enough to permit the use of free software, the very act of deploying it helps engender a culture that accepts mistakes and problems as natural.
Unfortunately, what stops people arriving at this happy state of affairs is a negative feedback loop that keeps them locked into both proprietary software and an exaggerated fear of failure. All the more reason for the UK government to help push open source into its offices to break that vicious circle. Let's hope this valuable new report encourages them to do that.