Something about this time of year makes me even more suspicious of spin than usual. So when I read Steve Ballmer's e-mailed message to customers and partners, I wanted to find out a little more about the research he used to defend his position against Linux.Ballmer cited a May 2004 Forrester Research study entitled "The Costs and Risks of Open Source" to hammer home the point that Linux projects take 5% to 25% longer to plan and deploy and that Linux training is on average 15% more expensive than Windows training. All true Julie Giera, a Forrester VP and author of the report, told me on the phone today. However, she also notes that her report's sample was very small because she simply couldn't find many companies that closely tracked their costs associated with Linux and who had been running Linux for more than a year.

Companies that start with Linux or other open-source applications from the beginning can save money over Windows or Unix, Giera says. The equation becomes more complicated when companies choose to operate a heterogeneous IT environment, as most do. "I can't think of any company over the past six months running Linux and Windows in the same environment who's saved money running Linux," she says. This will change, however, as Linux matures and companies build up open-source expertise.

Analysis and design of IT systems usually accounts for 20% of an IT project. Because it takes longer to do this using open-source software, a company's employees will be tied up longer on the project and the project's benefits will be delayed, Giera says.

Some IT managers say they can easily re-train their Unix programmers to manage Linux environments. Giera contends that good Unix programmers have always commanded a premium, and that time and money still factor into the equation when re-training. Open-source software is available largely in components today. "This requires a fairly sophisticated level of IT professional to pull together these things in an enterprise-class environment," she says.

Giera says the point of her report was not to skew opinion for or against Linux but rather to help companies understand the consequences of a shift in IT. "There isn't a big base of companies running Linux, so it's hard to tell the long-term cost versus the cost savings," she says.

All of this begs the question, what do you think?