IBM loves Linux. If this statement is a revelation to you, then you must be really new to the business. Either that, or you have just awakened from a slumber that would make Rip van Winkle's seem like an afternoon nap. IBM has the upstart OS running on every hardware platform that it makes—from the industry-standard xSeries to its massive zSeries—so it's patently evident that they are into the Penguin in a serious way. The question is, why does IBM find Linux so attractive? IBM has some mature and extremely powerful operating systems of its own (i5/OS probably being the favored pick amongst the readership of this Web site), so why would it embrace an OS that it doesn't own? Moreover, is Tux (the Linux mascot) likely to become a spurned lover in the future? In this article, I'll cover what, in my opinion, is the likely path that has brought IBM to this point and where the path will finally lead.

The Worst Business Decision of All Time

The roots of IBM's currently Linux strategy go back all the way to 1980, when it made arguably the worst business decision of all time: licensing MS-DOS from Bill Gates instead of buying it. The story is well-known and well-documented, so I won't rehash it here. Suffice it to say that this blunder occurred because the powers-that-be at IBM didn't believe there would ever be any large market for these new little 16-bit computers, so they couldn't see their way clear to own the OS. It's a mistake that has been haunting them ever since.

For a while, IBM and Microsoft (MS) had this nice, symbiotic relationship going. IBM built the hardware, and MS supplied the operating system. Once the market for the personal computer started flourishing, the brass at IBM couldn't help but notice the amount of money that they left on the table. They had put themselves in a subordinate position with Microsoft, and they knew it. For a computing powerhouse like IBM, being beholden to Microsoft had to be untenable. No doubt that the responsible executives had many sleepless nights over this; some probably suffered repetitive strain injuries reaching for the ripcords of their golden parachutes.

The Inevitable Divorce

While IBM had to make due with its PC hardware revenues, it saw its share of the market eroded by the many clone makers who were putting out better-performing hardware for much less cost than IBM was. IBM just wasn't used to dealing in a computer market like this and as a result was taking a serious hammering. At this time, it became apparent that the next big thing in the market was putting a pretty face on the command line—in other words, providing a graphical user interface. The IBM executives saw this as an opportunity to get back into the game, so they proposed a joint project with Microsoft to create a new operating system, which would become "Operating System/2" (OS/2).

Having struck the deal, IBM was no longer just a customer of Microsoft, it was now a partner, giving it serious leverage in the course that the OS would take. Unfortunately for IBM, the partnership didn't last long. IBM's idea was to build an operating system that would start with a fresh slate, ostensibly fixing some of the shortcomings of MS-DOS. My guess is that it was as much to ensure that their piece of the pie would be larger as it was to access more memory without the gyrations of EMS/XMS drivers. All Microsoft wanted to do was put the graphical face on the existing MS-DOS operating system, thus coming to market quickly and with minimal investment while at the same time keeping the lion's share for themselves. Like any marriage, these incompatible purposes caused the inevitable divorce. Microsoft eventually dropped its interest in OS/2 to work on its (in)famous Windows Operating System while IBM took full responsibility for its new baby.

The former partners were now direct competitors over the PC market, and for years, things in the PC OS market were fascinating. We had IBM working their magic with OS/2, even to the extent of attempting to wrest market share from Microsoft by including with it a "better Windows than Windows." Part of the separation agreement had IBM having access to the original Windows 3.1 source code, so that task was made easier. Anyone who had the pleasure of running Windows programs under OS/2 knows that the programs really did run much better than they did on a WinDOS computer. They ran faster and crashed less. Microsoft's response to this threat was to create a continually changing API that IBM had to incorporate, so its support of Windows programs always lagged the "real thing." I can only imagine what the two could have created had they actually cooperated instead of playing this game for all of that time.

We all know who eventually won this race, thanks to the facts that came to light during the United States v. Microsoft anti-trust suit. IBM's PC division was taking a hit on the price it paid for Windows to load on the PCs and ThinkPads it was shipping because it was shipping OS/2 as well. Microsoft just wasn't interested in that other OS getting out in any form, so it used its best weapon, and, well, you know how the story ends.

Moore's Law Kicks In

Once IBM officially put the nail in the coffin of OS/2, it found itself once again without a viable OS for its PCs, and it was once again a customer of Microsoft's. Things were OK, though. The other platforms that IBM owned exclusively were doing fine, thank you. For the customer space that IBM traditionally "owned," its pSeries, iSeries, and zSeries were earning well. Then a funny thing happened: Moore's law started to kick in, and processor speeds climbed, memory prices dropped, and disk drives began to reach incredible speeds with ever-increasing capacities. The tasks that earlier required IBM's bigger iron could now be done with the relatively inexpensive Intel hardware. More and more of IBM's customers started moving workload onto that platform, straight into the arms of Microsoft. IBM was once again in the unenviable position of being a hardware supplier—but this time, selling equipment that had a minuscule profit margin. While it was doing fairly well in the support and services market, the fact remained that its main competitor in that space not only provided services and support, but did so with first-hand knowledge of the product. This was not a good position to be in, by any stretch of the imagination. Something had to be done.

Enter the Penguin

I don't want to infer that IBM's fascination with Linux is a recent phenomenon, only that its public love-fest is. IBMers have been courting the penguin for almost a decade. I know this because of some discussions I had with Don Denoncourt back in 1998. I was taking a course about Java from Don in Atlantic City. We were supposed to take laptop computers to the course, and mine, naturally, was running Linux. While he routinely rebooted his laptop computer at lunchtime in an attempt to prevent embarrassing lockups during his presentation, I only booted mine when I entered the room in the morning. He and I struck up a conversation about Linux, and I'm proud to take credit for turning him onto the OS. But I digress.... Soon after our association, I got a call from him during which he pointed out a story about an IBMer getting Linux to run on a mainframe computer. He wasn't sure why one would want to do it, but I was. I knew that the mainframes could be partitioned into multiple systems, and we mulled over the possibilities of creating many instances of Linux on a single box. It wasn't too long afterward that he called again, this time pointing out a story in which a company did just that. I don't remember which company, and it isn't really important. The point is, IBM was dabbling in the Linux arena in the late '90s, when Linux was barely an adolescent in the computing world. Fast forward to today, and you can see that what may have been a skunk-works project within IBM has become a key point in IBM's computing strategy. Why is that?

Shrinking Market Share

If you want to know what kind of an impact a competitor's product is having on Microsoft, you need only pull out of your desk your trusty FUD-ometer and point it in the direction of Redmond. The higher it registers, the more likely that MS is threatened by the product. In the case of Linux, the FUD-ometer has been off the scale. We've been treated to an incredible litany of FUD from the Washington state giant. We have gone from "Linux is no threat" to "Linux is violating our patents." You know, then, that Linux is starting to hurt right where it counts: market share and the bottom line.

I call Microsoft's latest round of vitriol FUD because it's totally unlikely that Microsoft would ever actually go after anyone for these alleged violations; doing so would require them to disclose what patents are supposed to be infringed on. As I said in an earlier "Linux Letter" column, they wouldn't want to do that. It hasn't worked for SCO, and it won't work for them either. Besides, I don't think mutually assured destruction is what they have in mind. No, the only reason they're making these statements is to muddy the waters a bit, hoping to dissuade an overly cautious IT manager from implementing a Linux solution. That, and to get some corporations to roll over and pay protection money. Companies guilty of that will never see one dime of my company's money.

Now look at IBM's position: they have at their disposal a high-quality operating system that they don't have to take full responsibility for. Maintaining an operating system is an expensive proposition, and maintaining one that's proprietary for proprietary hardware is risky since both have to sell well to recover the investments. By embracing an operating system that can run on any of its hardware platforms and that has the added benefit of requiring them to maintain only certain pieces, they have put themselves into a superb spot. As a customer, I may not be able to justify the cost of an i5 for a particular purpose, but if I can use the excess capacity of the smallest i5 to run an instance of Linux, I may find the bean counters more amenable. On a larger scale, I may not be able to justify a zSeries for the corporation, unless I'm going to replace the thousand Intel boxes I have running Linux with one Z-box.

Besides the quality of the Linux code, look at the great publicity that IBM gets for free. They're now seen as a "hip" corporation, even by the youngest computer science/IT graduates. That hasn't happened for a long time, since most college grads are now being force-fed the MS Kool-Aid in school. Open-source is getting cool in colleges, and by extension, IBM is getting cool, too.

See what I'm getting at? IBM is now back in the driver's seat. They can now supply hardware that handles the mainstream workloads that Linux does so well, as well as clean up in the niches where their other platforms make sense.

That they can do all of this with an OS that's giving heartburn to arguably the biggest pain in their corporate rear end is just icing on the cake.

A Golden Anniversary?

So what does the future hold for the IBM and Linux relationship? Will they have a golden anniversary, or will this marriage end in divorce as well? That's a good question, and while I can't foretell the future, I can give an educated guess. To do so, you need to look at IBM's recent behavior.

IBM has been a superb member not only of the Linux community, but the entire open-source community as well. Actions speak louder than words, and IBM has been donating much time, intellectual property, and, yes, patents to the community. A simple Google search for "IBM and Open Source" will reveal contributions that you may not even know about. IBM has spent an incredible amount fighting the farce that is the SCO lawsuit, further demonstrating their commitment to Linux.

Still skeptical? You'll have to admit that Java skills are in big demand. Consider that IBM made an early commitment to the Java language and has been an incredible participant in that community. I would argue that Java wouldn't have gotten as far as it has without the legitimacy given to it by IBM's support on its various platforms. Would Java skills be so valuable had IBM not supported it? I think not. IBM has been at least as fervent, if not more so, about Linux. So I only see good things coming there.

I've read a lot of comments both pro and con from people about IBM and Linux. Many feel as though their favorite platform has been ignored as IBM's attention has been diverted. I look at it this way: I love the i5, and I want it to survive. If it takes Linux running on the platform for that to occur, I have no problem with it. We have all been saying how great the i5 is as an integrating platform. IBM's support for Linux on the i5 just supports our argument. IBM has embraced Linux for what I think will be the long haul. Embrace it yourself if you want to be in this game for the long haul as well!

Barry L. Kline is a consultant and has been developing software on various DEC and IBM midrange platforms for over 24 years. Barry discovered Linux back in the days when it was necessary to download diskette images and source code from the Internet. Since then, he has installed Linux on hundreds of machines, where it functions as servers and workstations in iSeries and Windows networks. He co-authored the book Understanding Linux Web Hosting with Don Denoncourt. Barry can be reached at This e-mail address is being protected from spam bots, you need JavaScript enabled to view it .