Linux Community Fights Back Against Microsoft Claims

Published: February 14, 2006

by Timothy Prickett Morgan

Microsoft has its "Get The Facts" campaign to try to deflect the attacks of the open source community in the ongoing battle for lower total cost of ownership in the data center, and now the Linux community has the "Get The Truth" counter campaign to try to demonstrate that Linux is better than Windows. Very little so-called research is unsponsored in the IT racket, and the new "Get the Truth on Linux Management" study released yesterday is no different.

This Linux counterpunch to Microsoft is sponsored by Open Source Development Labs, the place where Linux creator Linus Torvalds gets his paycheck, and Levanta, a company that has created a Linux provisioning and management appliance called the Intrepid M, and obviously has a vested interest, like OSDL, in promoting the idea that Linux is not a "high-touch," expensive platform to use in production running applications.

To that end, OSDL and Levanta commissioned Enterprise Management Associates to study the administration practices and costs at over 200 Linux user sites. EMA made no attempt to survey both Windows and Linux sites to obtain similar and comparable sets of information, which limits the usefulness of the data gathered. But the study's aim was to deal with the contention by various past studies commissioned by Microsoft that claimed Linux had higher system management costs. While the study's respondents were a little heavy on the technology, education, and service provider sectors, the distribution of company sizes was consistent with the entry server market, with almost half of the companies with annual sales of $5 million or less. The survey was the result of a combination of self-selecting group polled through the Web, a random sample of a list of several thousand CIOs and MIS managers who were polled by telephone, and 13 in-depth interviews to use as a control group and a ruler.

According to the study, the administrators polled are not deploying Linux by hand these days, but are using various tools to help them do the job more quickly. Some 75 percent of Linux administrators are using what the study called "sophisticated tools" could provision a Linux server in under an hour, and a third of those questioned said they could do it in less than 30 minutes. (Exactly what tools they were using is not clear, but I would guess is that Levanta was co-sponsoring this study to find out for marketing reasons.) About half of those without such tools--in other words, they did it by hand--said they could provision a Linux machine in under an hour and 20 percent said they could do it in 20 minutes. If you ask me, this seems to call into question the real value of these "sophisticated tools." The difference is not that great.

As for patch management, survey respondents say that they spend less than five minutes per server per week messing around with patches. This is mostly a testament to the vast improvements in the Red Hat Network and Novell's Online Yast Update patch support services. Patching a Linux server is now no more time consuming or difficult than running Windows Update on a Windows box. To put it bluntly, your Mom can do it, and, if she is like these administrators, she could even be clever and use Microsoft Operations Manager (MOM) to do it. The administrators also reported that supporting different releases or versions of a given Linux distribution had no impact on the time or effort in managing Linux systems. In fact, some of the respondents said that they had more versions of Windows at their sites than they did Linuxes. The administrators polled didn't spend a lot of time on viruses and spyware, either. About 88 percent of those surveyed said that they spend less than 10 minutes per server per week wrestling with virus and spyware issues and patches.

One of the reasons why Linux management is not as big of a deal as many in the Microsoft camp would have us believe is that companies who have Unix, Windows, and mainframe servers are now using tools that have been extended to manage their Linux servers. Linux is just another platform now. Moreover, storage area networks and network-attached storage are more or less agnostic when it comes to server platforms; they don't care what OS asks for data, so long as they have permission to ask for it.

The one area where the EMA study did gather a little Linux-Windows comparative data was administrator salaries. The distribution of salaries for Linux and Windows admins were nearly identical at the companies polled, and those admins who did both, by the way didn't fare better in the bargain. You get no premium in this market for having both sets of skills, apparently. There were slightly more Linux admins making more than $70,000 a year (22 percent) than Windows administrators (17 percent), but half of the Linux admins and nearly two-third of the Windows admins make less than $60,000 a year. The multi-skilled admins at the high-end are commanding a slight premium, and they are bringing up the averages a bit. This seems like statistical noise to me.

There doesn't seem to be a big difference in the number of systems managed, either. About 85 percent of the companies surveyed had up to 100 servers, and the Linux administrators managed an average of 15 machines compared to 12 for Windows administrators. When you add in the bigger sites, where companies have deployment tools and tend to prefer Linux (particularly ISPs running infrastructure workloads), Linux admins could handle 68 servers, compared to 32 servers for Windows admins. This sample of sites with more than 100 servers on site was not very large, however, so the differences in the platform-related data could be statistically invalid.

If you want to check out this detailed report for yourself, I am sure Levanta would love you to get it from its Web site.