The Linux server bandwagon

With IBM, Sun and Red Hat leading the way, Linux is poised to infiltrate datacenters on a much wider scale. Larry Seltzer looks at how such an exodus might affect Unix.

The mission is so loud and clear that if you were just introduced to the computer industry you'd think of IBM as a Linux company.

They've spent a billion dollars on Linux and they say it's paying off. What are they doing with it? Creative destruction.

Where's the money going? Check out a href="http://www.ibm.com/linux/" target="_blank">the IBM Web site and you'll see barely a mention of desktop systems. They simply have no interest in them, and in fact there are less for sale now than there were a few months ago. Stick a fork in the Linux desktop market.

IBM is very interested in using Linux to sell servers, though. This is the guts of what they're in business for--that and follow-on services. The fact that the server is free (as in speech) is a small point.

But IBM has a lot of product lines, and pouring a billion dollars into Linux has to come at the expense of some other products. At the top of the list must be the RS/6000 line and AIX, IBM's real  Unix implementation. AIX is certainly alive, and IBM sells some powerful servers (the pSeries) to run it.

If you're an AIX shop looking for modernization you can stick with AIX, but the modern way to modernize is by migrating to Linux. Applications aren't profoundly different (as they are in Windows) and so should be portable without a major hassle. Your administration tools and commercial applications are almost certainly available on Linux by now. Even IBM's TV ads tell you that they are more interested in Linux. I don't remember any basketball player named "AIX" on the team (perhaps he was traded to Cisco for "Router").

Red Hat is in the act too, with the upcoming update of Red Hat Advanced Server, what they call "the first enterprise-class Linux operating system." (Funny, I suspect that if you went to them last year they would have said that their products were enterprise-class back then.) The marketing emphasis here is on the cost advantages of Linux over mainstream Unix versions, the ones that we already know are "enterprise-class." In many markets consolidation is just the way of the world, and the Unix market sure looks like it's consolidating around Linux. (This is seriously ironic since Linux is a clone of Unix. What was wrong with the commercial Unix versions based on the AT&T-licensed Unix code base? The problem was that as open as they were, they were all essentially proprietary. Nobody made a living on them except as a vehicle for selling hardware. Now Linux will kill them all.)

It's even happening at Sun, where Linux servers--at least as low-end "edge" servers--are in customer demand according to the company. Of course, Sun would say that there's a marked difference between these servers and the Solaris boxes on which they make their living, and perhaps they're right. But it's my impression that marketing distinctions between product lines are a lot more significant--perhaps "credible" is a better word--to the vendor than to their customers. In other words, customers will have no trouble imagining that Linux can grow into the high-end of Sun's product line, displacing Solaris.

What happens then? Once all the Unix vendors are heavily dependent on Linux, they're in the same boat as Dell and Gateway and (gasp!) Compaq and HP with Windows servers. It's a commodity market. Customers will have a much easier time moving around from vendor to vendor based on price differences that today aren't worth noticing.

A few months ago I wrote a column where I argued that Linux threatens Unix, not Windows. This is the stuff I was talking about. Red Hat finally has the nerve now to say it out loud, but it's always been true.

Do you think Linux may overtake Unix in the server market? Share your thoughts in our Talkback forum, or send an e-mail to Larry.