Linux makes inroads into IT.
Microsoft has responded to the growing acceptance of Linux into information technology departments by trying to dislodge one of the most commonly held beliefs about Linux: that Linux is cheaper than Windows. Microsoft boasted of a study on the total cost of ownership (TCO) that claims, through some gimmicks, that Windows has a better TCO than Linux.
The study, conducted by IDC and funded by Microsoft, concluded that Windows is cheaper than Linux in four out of five areas studied: file serving, print serving, networking infrastructure, and security applications. The fifth area, Web serving, showed Linux slightly ahead.
The believable part of this study is that the initial cost of the operating system is not a significant factor in multi-year costs for running systems. For example, the study specified a Windows system running Windows 2000 for five years without an upgrade, thereby eliminating expensive upgrade software costs that impact Windows far more than Linux.
What is odd about this is that Microsoft has been trying to bring IT shops into a three-year or lower upgrade cycle. So the much-vaunted study is not realistic. Ironically, Linux systems show a remarkable ability to continue to run on older systems through many operating system upgrades.
The main cost factor difference, according to the study, was that Linux administrators are likely to cost more than Windows administrators, sometimes 50 percent more than equivalent Windows administrators. Even so, the cost differences in the networking category were slight, less than $1,500 for supporting 100 users on a network server. The other differences found, mostly thanks to salaries, that Linux systems were far more expensive to run, except for Web serving.
In a contrasting study, the Robert Francis Group found, for IBM, that Linux was cheaper than either Windows or Sun Solaris. This study showed that for serving Web pages, SPARC-based Solaris systems typically handle more users per system than Intel-based Windows or Linux systems. When comparing Windows to Linux, this study found that Linux servers were slightly more efficient than Windows, based on the same hardware configuration.
One telling conclusion from the Robert Francis Group study is that while Linux or Solaris administrators cost more than Windows administrators, Linux or Solaris administrators managed more systems than their Windows counterparts, reducing the overall cost.
In addition, the Robert Francis Group study lists a number of soft costs, such as the Microsoft requirement that each company maintain accurate records validating that they comply with Microsoft’s licensing terms. While not huge, this effort takes time and therefore costs something. This is also a cost not faced by companies that run Linux.