Originally appeared in MSI Magazine
March 1, 1999

Inmates Taking Over the Asylum

According to Microsoft testimony at its anti-trust trial, the emergence of Linux means real competition with Windows. In a memorandum that was widely leaked last fall, nicknamed “The Halloween Papers,” a senior Microsoft official described the potential threat from Linux, and then outlined ways that Redmond could coopt the movement in the bud through proprietary extensions, just like it tried doing with Java.

Linux is the little operating system that could. Its kernel is so compact that users claim adequate performance, even from all those 386 or 486 PCs otherwise destined for charity or the junkyard. Maybe a better metaphor is the Energizer Bunny. It just keeps going. UNIX programmers attest to the fact that it almost never crashes. Just lock it away in a closet, come back a year later, and do a reboot to soothe your conscience.

The issue of whether Linux offers real competition with Windows is in the eyes of the beholder. You may not necessarily run your ERP application on Linux, but there’s a good chance that your marketing group might be running the company website on a couple Linux 486s locked in some cabinet. International Data Corp. pegs Linux as the fastest growing server OS, with its share of the market more than doubling over the past year. This year, all of the major database vendors, with the obvious exception of Microsoft, promise Linux versions of their products.

With all the growing support, is Linux really ready for prime time? To answer that question, you must look beyond mere technology. Sure, you won’t run any large-scale mission critical applications on it, because Linux’s scalability has not yet been proven on anything beyond 2- or 4-processor machines. For the real answer, however, you need to understand the true nature of open source software, which is how Linux is distributed.

Like shareware and public domain software, open source software is freely available, often without charge. But, there are several key differences. First, users are allowed, if not encouraged, to improve the source code—but they are not allowed to keep their improvements to themselves. Enhancements must be made available to the rest of the community, creating bonds that are tighter than in most user groups. Linux is as much social phenomenon as it is alternative technology. As such, the Linux crowd is highly committed and emotional, because it is really everybody's baby. It’s not likely that SAP users have the same feelings for R/3.

But is this any way to run an enterprise software strategy? Open source advocates argue that the technology development model is superior. You get an almost limitless army of the world’s best software minds constantly ironing out any kinks that pop up. You don't have to depend on the kindness of your software provider to maybe deliver that critical patch. As one user told us, “If you’re willing to pay a million dollars for a support contract, you might get a fix. Otherwise, you’re probably better off getting a psychic fix.”

Even if we concede that the open source model is more robust than the proprietary model, there is still the question of whom do you call at 3:00 am. As we noted a few years ago in this column (see Keeping Up With the Joneses, July 1996), corporate software choices generally hinge on the question of available support, not technology. Otherwise, as we noted back then, for our desktops, we would still likely be using DOS-based databases rather than their clunkier Windows successors.

Yet, as nature abhors a vacuum, don’t be surprised to see the IBMs of the world mobilizing their service networks to finally assure IT administrators that nobody will get fired for buying Linux. That will clear the way for ISV support.

The next question is Linux’s economics. We conducted a spot survey of Linux costs. We found that, since most Linux applications crawled in through the back door for use in standalone applications, the cost picture currently resembles your typical pilot project. Although the licenses might be cheaper, support will be another story.

In the long run, Linux’s strength won’t its cost of ownership. Instead, it will be the degree to which it keeps UNIX vendors and Microsoft honest.