By Timothy Prickett Morgan
When IBM Corp first entered the commercial Unix server and workstation business back in 1990, the Big Blue seal of approval still meant something, and the very fact that IBM had entered the Unix market went a long way towards legitimizing the efforts of Sun Microsystems Inc, Hewlett-Packard Co and others. Those companies had already placed all their bets on commercial Unix eventually taking over the proprietary midrange and mainframe markets.
That hasn’t exactly happened, of course. Rather than Unix directly taking over the world, as Sun, HP and a host of others had thought it would, vendors of traditional midrange and mainframe kit have gone the other way and embraced Unix technology and open APIs; with the advent of the e-business, Unix’s progeny, various internet protocols, have also been enthusiastically adopted by proprietary software vendors, including the big daddy of them all these days, Microsoft’s Windows NT. Unix has taken over the world now that the Internet has, it just did it in a backhanded way that didn’t necessarily make it any easier for the big Unix vendors to make any money.
Some would argue that IBM’s participation in Unix has been problematic from both IBM’s point of view – since it has other bases to protect – and from the other vendors, which have been forced to adapt their product and marketing strategies to reflect what IBM is doing, both in the Unix market and outside of it in the server and workstation markets at large. Simply put, IBM is such a big organization and so integral to corporate computing that no matter what it does – no matter how well or how poorly – it makes waves.
Back in 1990, when I attended the original RS/6000 announcements and I was steeped in IBM’s midrange and mainframe cultures, I had no problem believing that when IBM said it was going to be the Number 1 or Number 2 Unix vendor in the world by 1995, it would do it. Given IBM’s technology roadmap and its commercial clout, it seemed hard to believe that upstarts like HP, which had a very respectable yet small server business, and Sun Microsystems, which had essentially invented the Unix workstation market and the notion that you could sell servers to link them together, could sustain the onslaught of an unleashed Big Blue. We were, of course, wrong, because that is not the way that it has worked out. And quite frankly, despite all the hoopla surrounding Monday’s RS/6000 announcement extravaganza – deadpan IBM marketers are apparently referring to this week’s announcements as the Sunscreen announcements and the S80 Condor server as Sunblock – IBM will have a long way to go to unseat Sun Microsystems as the world’s biggest Unix server vendor.
That’s not to say that the announcements (see Top Stories) aren’t significant, even impressive. But it is a far cry from the earth- shattering drama that IBM’s marketeers are trying to create surrounding the Sunscreen announcements. (One long-time IBM watcher quipped when he saw the announcements being billed as the most important since IBM entered the Unix market more than nine years ago that for them to be that important, IBM would have to drop out of Unix altogether. He had a point…) Still, you can’t fault IBM for trying.
Plenty of people wonder why IBM is still in the Unix business at all. The fact that it can have four different server lines with marketing plans and platforms that are equally incompatible ought not to be possible. But if Y2K has taught us anything, it is that businesses are as addicted to their applications, the platforms that run them and their employee skill sets (in a tight labor market, no less) as any junkie has ever been addicted to heroin. Mainframes are an addiction as much as they were ever a choice, and so are AS/400s and so will NT servers be as they become entrenched in the data centers and distributed computer rooms of the 21st century. The addiction holds true for RS/6000s because AIX is not the same as Solaris or HP-UX or Linux, and customers using RS/6000s have long-established relationships with hardware resellers, financing organizations and software suppliers.
Because it is not so easy for these customers to just ditch their RS/6000s, IBM is not really under that great a pressure to reconsider whether or not it should be in the Unix business. IBM has over 125,000 RS/6000 customers, and probably 90% or more of them buy its Unix servers (not exclusively, of course). IBM claims to have shipped more than 850,000 Unix machines since 1990. A fair guess is that about 15% of the installed base or 150,000 machines are servers, but over the past few years as the Unix workstation business has declined, perhaps as much as 30% or so of current ship rates are servers. RS/6000 hardware sales rival those of the AS/400, and will soon surpass sales of the venerable S/390 line. This is a good reason for IBM to stay in the Unix business, no matter how tough it is, no matter how expensive it is to develop and manufacture the PowerPC, Power3, Power4 and Power5 processors. IBM can’t afford not to be in the Unix market, especially with the AS/400 business also dependent on the underlying hardware that gets sold as RS/6000 Unix servers.
