4 min read

Windows: Cheaper Than Linux?

Open-source advocate concedes there's truth in study that shows Windows has lower cost of ownership than Linux.
A leading open-source advocate has acknowledged there's a lot of truth in a recent study that found that Windows has lower total cost of ownership than Linux.

The study, commissioned by Microsoft and performed by IDC, found that Windows was 11% to 22% less expensive to administer than Linux for network infrastructure, print serving, file serving, and security applications. Linux beat Windows only for total cost of ownership in Web serving, where Windows was 6% more costly than Linux.

Open-source advocate Bruce Perens said the study points up legitimate issues with Linux, which lacks the robust management tools available for Windows. But he said Linux is quickly catching up with Windows' total cost of ownership.

"I'm not conceding the point, but this is an area we need to do some work on," Perens said, looking back on Linux's 11-year history. "Look how far we've come in the past 10 years. Extrapolate how far we can go in a few years. We'll beat the pants off of Microsoft in TCO given enough time."

Linux developers have, until now, been focusing on basic functionality, and, more recently, adding a graphical-user interface to Linux for ease-of-use. There are proprietary Linux administration tools on the market, but Perens said he would like to see improvements to open-source Linux tools.

"We could get to a point where Linux and Windows are in a dead heat as far as total cost of ownership. That's what competition is about. We've given Windows a run for their money--they used to have no effective competition," Perens said.

Adam Jollans, a program manager for Linux solutions at IBM, said management tools are being developed rapidly for Linux. IBM Tivoli had only two management tools available for Linux a year ago; now it has 20. Computer Associates is also moving its management tools over to Linux. And upcoming Red Hat and SuSE Linux server editions are vastly more manageable than predecessors, Jollans said. The SuSE server is due to ship this month and Red Hat's latest version in May.

Perens added that the study leaves out some important factors. Linux will prove to be cheaper than Windows in the long run, he said, because Windows locks enterprises in to a single-supplier solution. "The way most IT houses buy software, they're following the addiction model." Also, using one Microsoft package compels users to buy others--for instance, Microsoft Exchange requires Outlook to function properly. "What happens if you decide to switch to some other mailer. Not only do you have to switch the server, but if you have 80,000 desktops, you have to switch 80,000 clients," Perens said.

He recommended that companies insist on using only applications that use publicly documented file formats and communications protocols, so that changing one application doesn't require changing a whole herd of other applications that depend on it.

The study surveyed more than 100 midsize and large companies, looking at the current total cost of ownership for Windows 2000 and Linux and projecting those findings out five years. The cost included the cost of software upgrades, procurement, and retiring and upgrading hardware, said IDC analyst Dan Kusnetzky.

"It's interesting to note that in all cases, the largest difference was staffing-related costs: administration, development, procurement, and retirement, and that is perfectly in line with studies we have done on many other operating systems and servers," Kusnetzky said. Staff-related costs tend to be 50 percent to 70 percent of the total cost of ownership for systems over five years.

Windows was found to be 13% cheaper that Linux for file sharing, 19% less costly for print sharing, 22.5% less costly for security, and 11% less costly for network infrastructure. The average company surveyed has 1,315 users.

While Microsoft funded the study, it was not biased in Microsoft's favor, Kusnetzky said. "We do studies for a large number of companies, and the methodology is the same," he said. "Microsoft is one of our subscribers. Most of the time, these studies are not made public. Microsoft funded us to do the study, they did not control what we asked and who we surveyed and what the results should be. If we did that our other subscribers would be angry, and so the damage we receive would outweigh any benefits."