Saving the System i: Fight Rather Than Switch
Published: December 4, 2006
by Brian Kelly
In 2005, Windows servers eclipsed Unix servers for the first time in terms of revenue, by nearly $200 million. According to IDC, over $17.7 billion worth of Windows servers and $15.5 billion in Unix servers were sold worldwide in 2005. Additionally, Linux took third place at $5.3 billion and mainframes slipped to $4.8 billion. With sales of approximately $2 billion in 2005, IBM's System i fell short of making the report.
How is it that the grandfather of what was once the darling small business system family in the world (the IBM System/3 and System/3X) could have slipped so far off the radar screen? The fact is that in the 1970s, the System/3 was the premiere small business machine, giving way to the System/32, System/34, and finally the System/38 in 1978. In 1983, the System/36 continued this rich tradition of providing no-nonsense, ease-of use computing for small businesses In those days, whenever someone visualized a successful and omnipresent (for that time period) small business system, it was inevitably an IBM model that started with the letters System/3X. The stark contrast with today's industry standards begs the question: What happened?
When the AS/400 was announced in 1988, IBM was poised to continue its success with a follow-up system to the System/38 and the System/36. This new machine was created with all of the necessary chutzpah to rival minicomputer vendors of the day, who had already begun to eat IBM's lunch. In the mid-to-late 1980s, the System/38 and System/36, even with all of their ease of use and advanced facilities, were comparably low on power and high on price. IBM had grown accustomed to gaining a premium simply for real ease of use.
From IBM's viewpoint at that time, DEC, HP, Data General, and Wang were the conspicuous villains blocking its accustomed path to easy success. The AS/400 was so perfect for its day that within a year or two after it emerged, it literally put DEC out of business. HP is the only survivor from that time period and even that can largely be attributed to its printer business and its recent friendship with Intel and Microsoft. HP defected to Intel chips--X86, X64 and Itanium, and has added AMD's Opterons, too--rather than continuing to fight as IBM did with its own chips. So again, what happened?
Just as Sony had the best VCR technology (Betamax) at the time of its introduction, yet was defeated by the empirically inferior Video Home System from Victor Company of Japan, a similar phenomenon repeated in the midrange system realm.
Let's examine what occurred with Sony and JVC. At JVC's Yokohama plant, Shizuo Takano, Yuma Shiraishi, and a team of engineers developed VHS technology. They followed a simple vision of the need for home video recorders and embodied it in their unique inventions. The first model using VHS was the JVC HR-3300, announced more than 30 years ago (on September 9, 1976, to be specific). Despite Sony being first (1975) and best (in terms of sharpness), JVC was able to entice 40 companies to adopt its format by 1984. Sony was able to acquire a mere 12 adoptions. For the consumer, there were far more units with VHS available than with Betamax. As new movies were always available in VHS and only occasionally in Betamax, consumer behavior cued Sony to acquiesce. By 1988, even Sony was making VHS recorders. VHS became the world standard for home video tape recorders--even though it wasn't the best.
The resonant moral of the story is that popularity and pervasiveness generally trumps design and engineering excellence. What good does it do for a company to have the best product if potential customers only see products made by competitors? As long as the competitors' products are adequate, a typical consumer, unaware of anything better, is going to buy the product. Even if the consumer is aware that there is something special about a product, if finding support for that product is more of a hassle than it's worth, then what confidence could they have that the superior product would be a wise long-term investment for them or their organization?
People eschew risk wherever possible, and most people go to great lengths to avoid weighty financial gambles. An unproven and expensive product with no marketing inertia gives a consumer a cold feeling about acquiring and using it. The unknown and unproven aspect of any product creates a risk factor that the potential buyer must overcome. It's common knowledge that Windows servers work. They are not unproven. They are in use in many AS/400 shops, often because IBM, in the 1980s and 1990s, believed its systems did not have to perform the functions that could be performed by PCs and PC servers because, after all, IBM had invented the PC as we know it.
Almost all executives know about Windows. They may not be able to differentiate a Windows server from a Windows client, but for the most part, they are not shocked and surprised when the beleaguered IT director decides to switch rather than continue fighting the AS/400 fight. Many wonder why it didn't happen sooner because, other than the director's pleas to keep the AS/400 or System i box, they would have never heard of the latter machines in the first place.
IT managers at Systems i shops face an uphill battle because many of the mixed technology staff know nothing about the system, whereas everybody knows something about Windows. Can the manifest destiny of Windows Server becoming the biggest source of server revenue in the world work to unseat loyal AS/400 shops from their favorite AS/400 systems? Indeed, and it is has been happening for years.
As much as IBM has turned its business over to its ISVs, its ISVs have been looking after Numero Uno rather than looking after IBM. Can anyone blame them? Show me an ISV with a solution that does not scale down to the mom-and-pop shop level and I'll show you an ISV who is about to develop and offer a "modern" version of a package that runs on Windows and its darling .NET middleware. Once an ISV develops that version, it quickly finds that the software is more popular and it can reach lots more customers. Why bother with something that makes your life more difficult?
Look no further than SunGard's HTE division for local government. If you want a modern inexpensive version of the software, as my client was told, "Go with the Windows version." HTE's System i group won't even call on the client if it is not sure they want a System i. The new account marketing is for new Windows clients, not new System i clients.
With IBM pushing ISVs to move to Windows to "modernize" the front ends of their applications, the ISVs are taking it a few steps further. Once they invest in the front end with a distributed database approach (JDBC and ODBC, among others), the next logical step is to port the server code to the Windows Server platform and .NET. Ironically, IBM's lack of a natural GUI on the iSeries, its own pressure with the "Roadmap," and its direct funding for receptive ISVs to modernize, gives ISVs the ability to sell wares on i5/OS, Windows, Unix, and Linux systems. As great a deal as it is for the ISVs, it is not such a deal for IBM and the System i. The ISVs reap the rewards and when they find out how many potential customers will buy their Windows software, they simply put the AS/400 version in the legacy closet, Windows front-end and all.
In an article in this newsletter called "The iSeries Goes to Town in Local Government," Mary Lou Roberts explains her discovery of three considerably large cities embroiled in a typical Windows/iSeries debate. She concludes by noting that in all three of these situations--where all are long-time (decades long) users of the AS/400 platform and have major investments in legacy applications--two specifically mentioned difficulty in staffing iSeries positions and submitted that as a reason for moving off the platform in the future. The third cited a need for IBM to get college students introduced to the iSeries.
Roberts suggests that the IBMers running the System i division may be sitting on a potential gold mine of local and county governments. But, at the risk of stating the obvious, these opportunities are definitively very local. To realize the potential, local governments have to be able to acquire programmers and system administrators to work on these systems beyond major commercial hubs like Chicago or New York, and in places like Cuyahoga Falls, Sioux Falls, or Wilkes-Barre, the burg in Pennsylvania where I live. There are myriad other smaller cities and towns in similar circumstances. It would be a shame to see the established installations abandon the OS/400 and i5/OS platform and its new prospects, turning instead to Windows-based applications, simply because IBM has not done everything possible to get the iSeries educational programs firmly planted within the community college system.
In every piece that I read about a successful System i shop, other than the casinos in Las Vegas and Atlantic City, there is this ominous cloud over the shop (that can be seen through all Windows) that while seeming stable now, hints at a looming thunderstorm. Is it manifest destiny that Bill Gates will rule the server world as well as the client world? Similarly, will Intel's chief executive officer, Paul Otellini, accompany Gates to the bank? Is this an ineluctable certainty? How did this happen and will it continue?
Earlier this year, an AS/400-loving network administrator in an AS/400 shop wrote to Slashdot to see if the gurus that hang out there could help in resolving the same issue that many AS/400 shops face today. (Which we told you about at the time in this newsletter as well.) Here is what he said:
"Our current ERP system has outlived its usefulness and we are in the process of selecting another package. Our present ERP system runs on an IBM AS/400, but there are those in the company who believe we should switch to something that only runs on Windows. My position is this: if we can find software that will meet our needs that runs on an IBM iSeries (new name for the AS/400) then we'd be certifiably crazy to move over to the Windows platform. A co-worker insists there are a ton of reasons to avoid the iSeries like the plague. I'm not trying to start a flame war, but if you were to bet your career on this issue, which side would you choose and why?"
After a lot of banter and opinion back and forth, one astute writer offered this simple but compelling analysis and recommendation:
"Back when you got your AS/400. . . whomever recommended it, made a safe choice; because nobody got fired for buying IBM. Today, Microsoft is in that same position. Micrsoft's ERP products for manufacturing are second to none. The price point for the software scales to your business needs and the X86 hardware is comparatively inexpensive. So you could bet your job on it!"
Microsoft is the new IBM? In the shrinking OS/400 and i5/OS ecosystem, most long-term IT managers have never really embraced Windows and do not care to. AS/400 people know and love the AS/400, the iSeries, and the System i, and they trust it implicitly. Yet each time that "package end time" arrives or "hardware upgrade time" comes about, these folks have to defend their platform like nobody else because IBM chooses not to. Each time, the result is the same; another percentage of these once loyal accounts go into the Gates/Otellini column, never again to return.
It is ironic that the same IT managers and programmers who will "never" consider Windows for production are Windows desktop users themselves. They can't help it. They do not want to turn the shop into a server farm experience and they protect the company as best as they can, yet the Windows operating system and the Intel and AMD processors sit on their very own desktops--except for the very small minority who might be using Linux on the desktop.
For the Windows gurus who never will hear about a System i, they are not badgered by continual requests to change platforms. Most reasonable people simply assume that everything runs on Windows anyway. Aside from their peers in IT, they may never hear anybody talking about an AS/400 or a System i. Therefore, all other things being equal (such as the applications still running and the farm environment staying up), it would take a mountain of reasons for a Windows user to investigate a System i. Why would they want to add work to their day? The fact is that the less the buyer knows about the System i (and yes, IBM does a good job of keeping this product a secret from the masses), the more likely they are to go with the product that they are using that is already popular and pervasive.
When analysts were commenting about the success of Windows server in the IDC report mentioned at the beginning of this article, some noted that they always expected that some day as Windows stabilized and the servers proved they could stay up for substantial periods of time, eventually Windows servers would overtake all other servers. In fact, in some articles, the analysts noted that if it were not for some launch delays it should have happened sooner. In other words, most analysts predicted that Windows would gain server dominance, which it now has, even earlier than it already gained it. Microsoft Windows server software on the X64 platform has now reached market dominance. How did they know? Why did they believe that it would happen? Why did it happen? And what does that say about the future?
And if you are from IBM you might be asking, "if everybody (that includes us) knew that Microsoft and Intel were going to steal our business at the server level just like they did at the PC level (remember DOS/Windows and the Intel X86 were only components of an IBM design and what was once an IBM monopoly), how is it that we just went ahead and let them? Is it manifest destiny? Is it too late? Should we continue our appeasement practice of helping Intel and Microsoft with whatever Bill Gates wants while we can until we are out of chips completely?" At a minimum, shouldn't somebody at IBM be asking that question?
The Germans have an intriguing saying, Machs Nichs. It essentially stands for the proposition that "it just doesn't matter." For most IT vendors--from Sun Microsystems to HP to Dell to Unisys--it really doesn't matter. These companies have moved to Wintel. But, to IBM with its competing Power chip family and its AIX, Linux, i5/OS, and z/OS operating systems on its midrange and mainframe servers, one would think that it ought to matter. Does it matter to IBM?
Some of us who have been looking at the IBM scenario for a long time continue to conclude that a house divided against itself cannot stand. Four competing brands in one server division is such a tough thing to manage that it's easy to see why IBM renamed them all a few years ago so they would all be eServers. Yet elephants, mice, camels, and zebra are different even if we refer to them all as animals. Despite a noble intent, calling all of the IBM servers "eServers" did not make them an easier job for IBM executives to manage, nor did it make it easier for IBM's potential customers to differentiate IBM's wares. In fact, if anything, it made it more difficult by having a misleading umbrella term encompassing so many genuinely distinct servers. Meanwhile "everybody's talking" about Intel and Windows. With all that talking, a whole lot of people acted in accordance with the zeitgeist and that's why Windows now runs the IT server world.
Does that mean that IBM should sell of all of its server divisions as it did with its failing PC division and concentrate on software? Why not get out while the assets can bring in a lot more than the $2 billion from Lenovo? Or can this problem of a shrinking server base coupled with the major competitor server somehow be managed? How so? Would it entail obsequious cooperation with Intel so that all of IBM's server machines get to have that familiar audio "doink" and Intel pays? So far, the only management that I have seen is that IBM is protecting its server flanks in its relationships with Intel, AMD, Microsoft, and the PC community. IBM is hedging its bets in case it cannot continue to keep its own server divisions flying. When was the last IBM ad you saw that did not include a little "doink" from Intel?
The biggest problem with this relationship is that it makes the players on both sides of the fight a little sick, like a chummy photo-op of Hillary Clinton hugging George W. Bush. It just doesn't seem right. Why would IBM be cuddling up to its biggest competitor if it did not think it would prosper if Intel dominated the Power platform? If IBM's PC server division can sell enough iron to make the bottom line look better, then why spare any concern for IBM using the Power6 or Power7, or Power8 or a future variant of its Cell processor in any of its own server products. IBM's services division would make a killing converting AIX and Z/OS and i5/OS boxes to Windows and Intel. Is that the plan?
Well, now if we sample what IBM's server or "system" ads are all about, we may get a clue. However, now that Lenovo owns the IBM client PC company, it seems that the only IBM advertising that has been coming from IBM lately, which, by the way has become very good, at the end features again that notorious Intel "doink." Where are the IBM systems? Why is it not OK for somebody who is not technically astute to be able to hear something about IBM that does not include the Intel "doink?"
Somehow a company that makes chips that compete against IBM's Power chips, with about half the revenue of Big Blue, controls the entire server advertising for almost all companies, including IBM. For somebody doing an autopsy on IBM's Systems and Technology Group, this little fact would scarcely go unnoticed. Every server ad is for Intel, yet the same cannot be said for IBM. It's no wonder that Intel chips are better known and loved than Power chips. I don't recall ever hearing or seeing an ad about a Power chip--anywhere, any time.
IBM does like the IBM name to be exposed so the world will buy IBM widgets of all shapes and sizes. Accordingly, the world will also buy IBM widget services and software. IBM must believe that the free ads from Intel are enough to get the IBM name out there and that the little "doink" at the end does not dilute the IBM logo. IBM, the company, still gets its scintilla of recognition.
Somebody at the top of IBM must know that the X86 and now the X64 server lines compete against its proprietary systems and operating systems. What a situation if you happen to run one of IBM's one of those lines of business at Big Blue. These poor IBM souls trying to get some business in a tough marketplace must accept as a given that the manager of the System x division gets all of the public advertising--even though it is funded by Intel. They have no prayer of stopping the Intel "doinks" coming from IBM advertising and they apparently cannot get a thin dime from the corporation to say a bad word in public against either Intel or Microsoft. Yet, Intel and Microsoft are killing IBM in the marketplace. It is not only unbelievable, it is a losing strategy to advertise your competitors' products. One doing so might be well advised to check the standings and check the IDC report; IBM is losing market share all the time in its "system area" while Windows and Intel are gobbling it up to beat the band. (However, IBM gets to remain the top revenue dog across all server types, which is where the bonuses must come from in Systems and Technology Group.)
IBM can change that and hopefully one day, it will. However, that is not the big reason why nobody other than its owners seems to know or care one iota about the System i. I am not intimately acquainted with the finer details behind the Lenovo deal, but one thing is for sure today in IBM: There are no client systems. Just as Intel "doinks" seem to please IBM executives by virtue of IBM's slim concomitant mentioning, those little desktops and the best laptop in the industry no longer say IBM. So, overall I would say that the average Josephine sees even less IBM with all the "doinks" than when the laptop next to hers at the airport was a ThinkPad.
IBM is cleanly out of both the commodity and the non-commodity client PC business. That also means that there are no more Intel "doinks" left to be heard on the client side. I think what's good for the client is good for the server. I would suggest that IBM realign itself with Lenovo or some comparable entity and create an independent company with its own stock that sells IBM PC servers. As a carrot for Lenovo, give them a license for the IBM logo on all of these machines. As a formula for success for IBM, do not send any PC matters to Armonk or Somers for resolution. Let Lenovo run the client and server business, but brand it IBM.
Were IBM to adopt this strategy, it can clean itself out of the "doink" business completely and thereby could concentrate on its own system lines. Perhaps it would actually be competitive against each and every "doink" that you and I ever hear again.
Some suggestions: Let all new IBM messages have no sense of homogenization. Let each system line be able to differentiate itself and when conflicts arise, let higher IBM managers be empowered and resolved to make the best decisions, giving the division managers the benefit of the doubt. Keep Intel and Microsoft managers and executives out of the IBM system decision process. They've already done enough to suppress the growth of IBM stock.
Further, permit the heads of the all-IBM system divisions to take their spirit and their creative thinking to the job with them every day. They'll be able to think of "what if" scenarios without worrying about getting "doinked" on their way to the boss's office. These IBM managers will eventually believe they can be successful rather than having to accept the mediocrity of an Intel-dominated IBM headquarters. One of these managers in turn would naturally be free to make the System i successful. And it would be fine with me if that general manager was Mark Shearer.
In next week's issue, part two of this article: How IBM's System i can make its comeback.
Brian Kelly is an assistant professor in the Business Information Technology program at Marywood University, where he also serves as the System i technical advisor to the IT faculty. Kelly has developed and taught a number of college courses in the IT and business areas. He has an active consultancy in the information technology field. Kelly is the author of 27 books and he has written numerous magazine and e-magazine articles about current IT topics. He is a frequent speaker at the COMMON National Computer Users Group and he has been a featured speaker at numerous technical conferences and users group meetings across the United States.