I haven’t been to enough COMMON conferences to judge what a typical show is like, but I can tell you that, despite IBM not announcing the I-Star servers in February just prior to the conference (as many analysts and customers were led to expect), it was still pretty lively and interesting. According to COMMON President Bob Cozzi, some 4,000 AS/400 users, Business Partners, and IBMers showed up either to hear or to talk about plans for AS/400- related products. Hundreds of vendors were on hand to talk about what they were up to, and the staff at Midrange Computing talked to as many of them as was possible to enhance our coverage of midrange players, both big and small. We even set up the Virtual COMMON section on our Web site (www.midrangecomputing. com/common) to give those of you who could not travel to San Diego an idea of what went on there. The site includes product announcements as well as transcriptions of the keynote and opening- session addresses. In this month’s “Midrange Insights,” I’ll give you an overview of what was said in those addresses and get into some of the juicy rumors that were running around the conference.
The keynote address was given by Robert Tipton, a guru from IT consulting firm marchFIRST (formerly Whittman-Hart). Tipton gave a presentation that centered on the often-troubled history of the AS/400 within IBM (the AS/400 is extremely reliable, but IBM’s internal treatment of the platform has had its ups and downs) and how that history contrasts with other companies that have had better success in establishing their technologies as mainstream products. But before he did that, he put IBM on notice. “I want to say a couple of things to IBM,” Tipton said before he got rolling. “I want you to look at this session as putting you on notice at this COMMON meeting. I want you to feel uncomfortable. I want you to know that there is a different way that we can look at the AS/400 than what we are getting from IBM. So if there are IBMers here, I want you to feel a little nervous. And for the users here, I want you to feel passionate about this platform.” Whether or not his speech actually made AS/400 partners and customers passionate about the AS/400 or angry at IBM is debatable. But it was certainly enlightening and entertaining.
The impression that Tipton’s speech gave was that what separates the AS/400 from the Apple Macintosh, Microsoft Windows, or the Sun Sparc server and Solaris operating system is not technology but personality. “The interesting thing about these guys,” he said, meaning Apple’s Steve Jobs, Microsoft’s Bill Gates, and Sun’s Scott McNealy, “is that they are passionate. They are driven. They are maniacal about what they want, and no one
is going to move them off of that. They know exactly where they are going. Furthermore, each of these men was involved in the inception of the products that their companies sell today. Love them or hate them, they inspire fierce loyalty among the people who know them, and they have changed forever how we look at technology: Steve Jobs, the way we look at interactive systems; Bill Gates, the portable, mass-marketed software; Scott McNealy, the network is the computer.” The AS/400 seems to be missing its maniacal genius, although, in Frank Soltis, the AS/400 certainly has a genius.
The day after Tipton gave his keynote, three IBMers responded to Tipton. Paula Richards, an AS/400 application technologist, started the opening session with a long list of AS/400 customers doing e-business on their AS/400s, explaining rather briefly what dozens of AS/400 shops are doing with their AS/400s. Her presentation centered around a simple idea that IBM has been trying to drum into the heads of new and prospective AS/400 customers since it coined the term e-business back in August 1997 with the AS/400 Apache server announcements. The idea is that AS/400 e-business is not an oxymoron.
Richards was followed, in the opening session, by Malcolm Haines, AS/400 minister of public enlightenment, and Frank Soltis, the AS/400’s chief architect. Haines took the stage first, and the essence of what he said can be summed up in a few quotes. “Everyone here is an AS/400 customer, and that means everyone here loves AS/400,” he said. “We know that the product is very special in that it evokes huge devotion on the part of people who know this machine and base their business on this machine. And that alone should be enough evidence for people who think otherwise to say that AS/400 is quite unlike any other machine that IBM makes or that anyone else makes and that this is not to homogenize all servers and to pretend that all servers made by everybody are a kind of commodity.”
IBM’s Soltis followed up with a brief speech, explaining that the AS/400 has had plenty of technologies that are only now being adopted by other server manufacturers. Some of these technologies are now decades old, as far as the AS/400 and its predecessors, the System/36 and the System/38, are concerned. “Working on the AS/400 is one of the most rewarding and most frustrating things that you can possibly do,” Soltis said in his opening-session speech. “When we announce a new, great technology that nobody else in the world can possibly do, you folks yawn. You say, ‘That’s the AS/400. I expected that.’” This got a lot of laughs.
Soltis got even more laughs during a presentation later in the week when he went through the ups-and-downs history of the AS/400 Division and its predecessor, the General Systems Division, and how difficult it has been for that part of IBM to get what it needs to compete effectively in the midrange market. While IBM and I often don’t agree on the interpretation of AS/400 history, I think most IBMers would concede that, more than the AS/400 press and analyst community, IBM itself has been at fault for not adequately putting forth the AS/400 as a superior solution as it tried to protect its various product lines and surf the waves of popularity—first UNIX, then Windows, and now UNIX, Windows, and Linux together—to boost other server lines. I would argue further that, by keeping AS/400 acquisition costs high relative to other platforms and, therefore, keeping the customer and independent software vendor (ISV) bases small compared to much larger bases such as the UNIX or Windows areas, IBM has given away billions of dollars to its competitors who would not have had such an easy time competing against a low-cost AS/400 that offers superior reliability and scalability. Exactly how IBM can behave in such a way, essentially giving away markets to its competitors, is beyond explanation. But I can tell you this: If Soltis had been CEO of IBM for the past decade, the AS/400 would have long since been an OS/400 and UNIX box, and Bill Gates would still be trapped on the desktop, except for some print-and-file serving with his Windows environment.
Rumors du Jour
Maybe they should change the name of the COMMON conference to AS/400 Rumor Fest
2000. Standing on the COMMON exhibitors floor, you could almost see rumors moving from person to person like a wave at a football game. If you are a rumor monger and a wise guy like me, there isn’t a more fun place than COMMON. Plenty of rumors whipped around at the show. Some of what I heard had the ring of truth; some had the clank of misunderstanding or outright falseness. There isn’t enough space in this column to get into all of them, but let’s go over the big three.
The first big rumor at COMMON was that Tom Jarosh, general manager of the Mid-Market Server Division, had moved on to a new job at IBM as so many of his predecessors have done in the past decade. The scuttlebutt was that Jarosh had a big meeting with IBM Chairman Louis Gerstner in New York while the show was taking place in San Diego. As far as we know, Jarosh is still in his job and happy with it. He ought to be, with effective control of over 99.9 percent of IBM’s commercial accounts. What is true is that Jarosh has brought Debra Thompson, a longtime AS/400 marketeer and evangelist who has been working in the RS/6000 organization, back into his organization with the specific job of sorting out IBM’s midrange platform conflicts. It is likely that COMMON attendees, hearing that she is back working for Jarosh, jumped to the conclusion that she had been given the top job. That may happen eventually, but it hasn’t happened yet.
The second big rumor at COMMON was that Dell is looking to buy the AS/400 or RS/6000 lines from IBM. High-level executives in the AS/400 section of the Mid-Market Server Division scoffed when I asked them about the rumors that Dell would buy the AS/400 line from IBM or was interested in buying the RS/6000 line to use it as a high-end Linux server. Even if the idea is absurd, it makes sense that Dell might be in Rochester trying to figure out if IBM’s RS/6000s or AS/400s would be appropriate platforms on which to run its own burgeoning direct PC and server business. One of the dirty little secrets in the PC market is that Dell was unable to move from its homegrown enterprise resource planning (ERP) applications, which just happen to run on Compaq’s Tandem Himalaya fault-tolerant servers. Dell attempted to implement the SAP R/3 suite a number of years ago after Compaq bought Tandem, which was just before it bought Digital Equipment Corp., and that project completely failed.
Big Blue is Dell’s technology partner by virtue of Dell’s multibillion-dollar parts deal with IBM’s Technology Group. (Jim Vanderslice, who started IBM’s Technology Group, is now working as Michael Dell’s second in command, so Dell has very high-level access to IBM and its technologies.) It is possible that Dell would use this OEM relationship and Vanderslice’s influence to look at the PowerPC Pulsar, I-Star, and S-Star chips as well as the Power4 chips due in the second half of 2001. It is even possible that Vanderslice is trying to get Dell to rethink its server strategy. But Dell is a Microsoft Windows NT/2000 and Intel PC server snob, and that is what it sells to customers. So Vanderslice will have to do some fast talking to get Dell to change its tune and create a high-end Linux business or a Power4 Win2K server business. Dell has been a laggard when it comes to promoting Linux as an alternative to Windows NT/2K, and it continues to drag its heels. Dell had the perfect opportunities to buy either Tandem, Sequent Computer Systems, or Data General, the latter two offering decent UNIX servers that are based on non-uniform memory access (NUMA) clustering and supporting Windows NT/2K, which could have bolstered Dell’s high-end server line. Being one of Tandem’s biggest customers, Dell should have seen that it was a potential target for competitors in the server biz. Having lost Tandem to Compaq, Dell didn’t learn. The company totally snoozed and lost out to IBM (which bought Sequent) and EMC (which bought Data General). Dell does not want a UNIX server line, even though many in the industry (including myself) have said it had better do something to get a more serious environment than Windows NT/2K for its high-end customers. So the odds that Dell wants to buy the AS/400 or RS/6000 line from IBM are not very high. Dell may want to resell the line if it thinks it can make money,
but I doubt that IBM wants Dell to take the AS/400 and RS/6000 lines direct, because doing so would disrupt their channels.
The other big rumor at COMMON was that Microsoft wants to run Windows on AS/400s and RS/6000s. In late 1998 or so, Microsoft apparently learned that IBM was going to support AIX applications in the AS/400’s Portable Application Solutions Environment (PASE). Running multiple operating systems on AS/400s is not a new idea, and many within IBM Rochester had been trying to get the go-ahead to create a PowerPC- based server line running AIX, OS/400, Windows NT, Mac OS, and other operating systems since the mid-1990s. Talk of such a universal server—indeed, what could really be called a magic box and not just be marketing—has been going around since IBM started using PowerPC chips in its server lines. This, of course, never happened.
Windows NT was originally conceived to be a multichip operating system. The idea was to take the Windows server environment mainstream by running it on all the big server chips: Intel x86, Digital Alpha, IBM PowerPC, and Silicon Graphics (SGI) MIPS. If Bill Gates could have killed Sun Microsystems’ CEO, Scott McNealy, he would have gotten Windows NT running on the UltraSparc chips, too. For whatever reason, Hewlett-Packard (HP) was not interested in running Windows NT on its PA-RISC chips, and, subsequently, HP has opted to run its UNIX as well as Microsoft operating systems on the Intel Itaniums, which HP actually helped Intel develop.
Mostly for marketing reasons, Microsoft and IBM dropped support for Windows NT on the PowerPC chip in late 1996. This was only a few months after IBM finally got PowerPC/NT workstations out the door, which was a bit of an embarrassment. Just before that, despite protestations from SGI and its chip partner, NEC, Microsoft decided it would no longer develop Windows NT for SGI’s MIPS chips. And late last year, struggling Compaq pulled the plug on Windows NT for its Alpha processors, leaving customers using that machine with its Tru64 UNIX as the only operating system option. That leaves Windows NT/2K running on Intel x86 and IA-64 processors. Microsoft is back to having only Intel as its platform partner.
Microsoft has apparently made two proposals to IBM concerning the AS/400 and RS/6000 lines. Having heard about IBM’s PASE plans for the AS/400, Microsoft approached the Rochester team and proposed that IBM develop a Windows NT runtime environment for OS/400. While this would be technically easy—remember, the 32-bit side of the PowerPC chip can run Windows NT already—sources within IBM who are familiar with the talks between Big Bill and Big Blue say that Rochester nixed the deal because Microsoft would not spill its Windows NT/2K development plans or source code so OS/400 programmers could see what was coming down the pike and be prepared to ship Windows NT/2K support more or less concurrent with Microsoft’s ship dates for releases. Rochester said “no way” to the NT implementation of PASE. The Rochester OS/400 programmers went with AIX because IBM owns it, and they can get the source code, beta code, and development road map from the Austin AIX team. And later this year—when AIX and its spin-off for Itanium chips, the Monterey/64, support Linux binaries through APIs and application binary interfaces (ABIs) being woven into AIX—PASE will be extended to run Linux binaries natively on AS/400 hardware. IBM Rochester is willing to support Linux within PASE because it can get source code and participate in the Linux development process.
Last October, Microsoft made another trip to Rochester to make yet another proposal. This time around, rather than playing the PASE game, Microsoft wanted to port Win2K to the Power4 chip that IBM is developing for AS/400 and RS/6000 servers. Apparently, Microsoft has agreed to do all of the work. In this scenario, Win2K would run in an OS/400 or AIX dynamic partition (yes, logical and dynamic partitioning as well as integrated NUMA clustering support will be in the Power4 servers for both the OS/400 and AIX flavors of the boxes), although it would, in theory, be possible to have Win2K just take over the whole machine. The problem with having Win2K run directly on Power4 servers is that it negates one of the key differentiating factors IBM uses to sell AS/400s and
RS/6000s: scalability. If Windows scales as well as OS/400 and AIX, it makes AS/400s and RS/6000s harder to sell. This kind of thing makes IBM marketeers break out in cold sweats. Odds are that Win2K for Power4 will not see the light of day, because the Netfinity people within IBM will want to push Intel-based boxes, not Power4 servers. From my perspective, native Windows NT should have been announced with the Apache AS/400 and RS/6000 servers back in August 1997.