Mad Dog 21/21: Wintel Overture
April 4, 2016 Hesh Wiener
In 1981, IBM caused a tectonic shift in computing by announcing its PC. IBM ultimately failed in the PC business, but it spawned an industry segment that still stands on the PC’s two pillars: Intel chips and Microsoft software. Thirty-five years later, however, most clients are mobile, with ARM chips and ‘nix variant software. Meanwhile, enterprise Wintel and Lintel have been enlisted by Microsoft for a fresh assault on the glass house archipelago. To avoid a debacle, IBM must reinvent its Power and mainframe platforms on the ground and in the cloud.
IBM appears vulnerable, adding to customers’ anxieties. This vulnerability is visible in the broadest measure of IBM’s situation in the marketplace. The company has suffered diminishing revenues quarter after quarter for 16 consecutive accounting periods. This is the entire period during which the company has been led by Ginny Rometty, except for her first quarter of service, which critics suggest reflected results locked in by her predecessor, Sam Palmisano, before Rometty’s influence could be fully felt.
On the other hand, IBM points out that its financial results reflect not a simple decline but rather a transformation as it shifts its activities away from hardware manufacturing and toward cloud computing and services that it hopes will provide it with fresh growth. For now, the company’s executives seem to be telling investors and customers that some legacy operations are indeed shrinking faster than its new strategic initiatives are growing, but this is only temporary. IBM’s revised strategy is beginning to get traction, and before long improved revenue and earnings will become visible.
Still, worriers are understandably concerned that some very substantial portions of IBM’s business are ultimately doomed, unable to produce the huge revenue and profit required to sustain the appetites of a beast that, notwithstanding the past few tough years, still enjoys annual intake north of $80 billion.
The tragic history of the IBM PC business might be an episode for investors or customers to view as a story portending the company’s likely future. But for a more rounded and considerably brighter picture, observers, investors, and enterprise users might want to examine IBM’s history in two other matters: the storage-oriented system technology that became its midrange computing family and the comprehensive large systems phenomenon that began with the deployment of the System/360 and continues with its evolution into products currently called the System z series of mainframes.
In 1958, some eight years before it introduced what would become the company’s mainframe line, IBM began offering a standalone computer built around a 5 million character magnetic disk storage system called the 350. The processor, named IBM 305 RAMAC, was created using vacuum tube logic circuits and other technologies of the mid-twentieth century. It weighed a more than a ton and required kilowatts of power. It may seem almost laughable compared to today’s computing devices from phones and calculators on up, but at the time it was a long leap ahead of alternative data processing hardware, which was for the most part a collection of electromechanical machines that processed data on punch cards and produced output as hard copy, commonly on fanfold papers pre-printed with green bars to make the results easier to read.
A year later, in 1959, IBM announced what would in retrospect become its first successful midrange system, the 1401. This computer used transistors rather than vacuum tubes, and its circuits were built on cards that had individual transistors, each inside a little metal can, rather than integrated circuits packing transistors printed on silicon wafers.
Big Blue cranked out something like 12,000 of these IBM 1401 machines, which remained in use well into the 1980s in south Asia, long after they had pretty much disappeared in North America and across Europe. The 1401s had unique disk drives with had better price/performance than the predecessor hardware of the RAMAC. Still, at the time disks were pretty exotic, and the most popular storage medium used with the 1401 was magnetic tape.
The 1401 ultimately displaced the relative handful of RAMAC systems in the IBM customer base. It coexisted in the IBM product line with larger, faster and more elaborate computers such as the 7090 and 7094, systems loved by the scientific and engineering institutions of the day. But it was doomed, as far as IBM was concerned, when Big Blue began offering the System/360 in 1964. The 360 line included small models designed to replace the 1401, mid-sized models that were upward compatible with the entry products, and, eventually, much larger models that had greater performance than the 7094.
Despite its relatively short life during the dynamic (for computer manufacturing, at least) 1950s, the RAMAC had huge influence on IBM’s corporate computing culture. During the 1970s, when Digital Equipment Corp minicomputers were winning business in situations that didn’t warrant the cost (or capability) of the IBM 360 and, later, 370 machine, IBM remembered that it once knew a lot about making small business systems. The ghost of RAMAC remained invisible as IBM built up what would become known as its General Systems division . . . until the creation of the System/38. This progenitor of the AS/400, iSeries, System i and now Power Systems running IBM i was built around the idea that information storage technology should define the hardware, software, and in fact the whole conceptual architecture of a business computer.
But, sadly, IBM has, during the past few decades, has paid too little attention to its scientists and engineers and too much attention to its marketing specialists. Consequently, it has lost its appreciation of the value to customers provided by the DBMS system and related technology at the heart of the System/38 and its follow-ons. IBM seems to think it can get away with its inattention to technology because despite its intellectual laziness it has been pretty successful competing with Oracle, the other mature giant of enterprise DBMS.
But for too long IBM has ignored the growing enterprise customer appreciation of Microsoft’s SQL Server and its middleware mishpucha. After all, Microsoft’s DBMS technology was, until now, tied to Windows Server, and Windows Server, in IBM’s apparent view, doesn’t pose much of a threat to IBM’s mainframe or Power environments in the big ticket enterprise accounts. Sure, lots of IBM’s enterprise customers like small- and mid-size systems using Window Server systems, including the ones using SQL Server as well as the Exchange email system. But mainframe shops that want an operating system they can count on for really massive jobs turn to z/OS or one of the fortified Linux distros that seem to run well on the IBM z platforms. Similarly, Power shops, even big ones, mainly love Unix, sometimes love IBM i and, lately, also are learning to love Linux. But mainframes and Power servers cannot run Windows Server. The result has militated against X86 boxes in the glass house.
Unfortunately for IBM, that has changed.
In March, Microsoft said that it was porting SQL server and lots of related software to Linux. Now Microsoft isn’t talking about mainframe Linux or Power Linux here, it is talking about X86 Linux. So it might mean that if a customer wants to stick with IBM for MIPS but try Microsoft for DBMS technology, it can’t just run the jobs on the Power or mainframe boxes. Chances are, SQL on Linux on IBM proprietary hardware will take a little learning; it won’t be exactly the same as running SQL on Linux on X86.
But these same customers might work with IBM as a cloud host. IBM can provide Power or mainframe service in the cloud, but it believes its future in the cloud is going to be mainly one based on X86 technology. That is why Big Blue has pumped billions into X86-based cloud services and has vowed to compete directly with the other biggies in the cloud game: Amazon, Google and Microsoft.
At the very least, just the way decades ago a computing executive that kept an Amdahl coffee cup in view might find that IBM was happy to become more aggressive in pricing, an enterprise customer with a workload that might run pretty well with the help of SQL Server ought to think about keeping some Microsoft manuals in plain view when IBM’s sales suits come to call.
IBM probably hasn’t yet figured out what is about to happen and might not until Microsoft has eaten so much of Big Blue’s lunch that IBM’s corporate tummy is rumbling. But eventually IBM, however dimwitted, will catch on. Fear of death can do that. One way or another, IBM is going to have to compete with Microsoft on Wintel and Lintel in the glass house and in the cloud because Microsoft is also running scared.
Microsoft has seen that the marketplace is treating Windows 10 client software like crap, or worse, like Windows Vista (and even has trouble getting its own productivity apps to work properly) but that its server software and, more importantly its DBMS and other middleware, is still held in pretty good regard by serious customers. Microsoft knows it cannot persuade anyone that Linux is a bad idea (even when it turns out to be a bad idea) and that it can do a great job selling enterprise software into the Linux base, particularly if it can stay a step or two ahead of IBM and Oracle when it comes to pricing.
It doesn’t matter right now whether DB2 or Oracle have features that are better than those in SQL Server. If 95 percent of the new apps that enterprise users want to create can be implemented properly with Microsoft software that costs a lot less than the alternatives, the old timers in the enterprise market are going to get clocked.
Because IBM (and Oracle) folk read the same news media everyone else does, Microsoft’s expanded ambition is hardly a surprise. IBM retains one advantage over Oracle: lots of customers love IBM hardware; a much smaller base loves Sparc stuff. But Hewlett-Packard Enterprise and Dell, among others, are happy to sell and support X86 boxes built with enterprise customers in mind. The enterprise class X86 vendors don’t have their own systems software so they are obliged to sell Windows or Linux. And they are pretty good at proving (or trying to prove) that their X86 offerings give more bang for the buck than a competing Power or mainframe platform.
IBM will often be able to hold onto its base but it cannot win in every case. It can, for a little while, find ways to churn the eroding base, but doing so leads to resentment on the part of customers even as it increases revenue. It looks like IBM is going to have to come up with some new wrinkles to keep its glass house business from slipping away faster than it can build up its new strategic initiatives.
Still there is a big open question. Does IBM have a viable vision, some strategy or group of strategies that will help it remain strong, healthy, and deserving of customers’ trust even if its legacy operations continue to weaken?
IBM has to be a lot smarter now than it was during its PC adventure. In the 1980s, IBM was able to shake off the antitrust constraints that had limited its choice of tactics. But it overplayed it hand, and in the 1990s, customers that were fed up with IBM’s bullying found ways to reduce their acquisition of IBM mainframe equipment. At the same time, Big Blue was having difficulty getting a very happy System/3X base to embrace its designated successor, the AS/400. IBM seemed to regain its effectiveness during the 1990s under the leadership of its first outside CEO, Lou Gerstner.
Gerstner never seemed to understand the essence of IBM’s prior success in midrange or mainframe products. Instead, he studied customers’ budgets and came up with the brilliant insight that users were spending at least as much on internal software technicians as they were on purchases of software or hardware products. Gerstner got IBM to launch a services business that allowed user companies to shift much of their staffing and consultant expenditures to IBM, which was able to provide services that delivered excellent value and quality. A key aspect of Gerstner’s strategy involved the develop of ways to get most of IBM’s services work done in India, where talent is high and where, compared to the USA and Europe, wages are low.
Gerstner’s concept seems to have pretty much gone as far as it could. IBM’s services business isn’t growing by leaps and bounds as it did twenty years ago. On the contrary, it is shrinking.
But Gerstner’s failure to understand the systems business was an error Big Blue seemed to be able to afford. That is not the case when IBM lost track of the plot in the segment that started as a PC business.
IBM thought the descendants of its PC would be similar machines like the PS/2. Well, that wasn’t the case at all. The PC and PS/2 are just what’s left for an office after you take the typewriters away. The real children of computers one can lift up are other digital storage, presentation, and communication devices. This is what Apple saw that IBM did not. First Apple took on the unconnected world, killing Sony’s Walkman stuff with its iPod. Then it saw that connectivity was becoming cheap and ubiquitous and also that it had at least three forms: the wired Internet, radio telephony, and Internet over WiFi. Apple built a download store that revolutionized music publishing and distribution and later transformed the software business.
With a superior delivery system, Apple, despite or very likely because it had no prior experience in the phone business, was able to create the miraculous iPhone and later the iPad. Apple is a leader but it is not alone. Amazon’s interpretation of the same business environment includes Kindles and Alexa, the former focused on media content, the latter on a broad spectrum of goods and services. These two companies’ visions of evolving clients are the actual descendants of the PC.
IBM still doesn’t quite get it. It is building small numbers of exotic and expensive Watson boxes when it ought to be building tens of millions of hundred-dollar corporate helpers. It’s not that Watson isn’t impressive. It’s a great bit of software engineering. But as a business, how is it likely to fare compared to the iPhone or one of the phone’s Android rivals? How is Watson as a service or as a stand-alone product going to sell compared to the iPhone or iPad or Kindle Fire or Alexa? Even if Watson turns out to be a winner rather than an expensive public relations maneuver, the big money is going to be made on the devices and services that get people to Watson, not on Watson itself.
No wonder Microsoft thinks it can successfully invade the glass house. All it has to do is wait for senescent IBM to be distracted by the challenge of fishing around in its pockets, hunting for its corporate AARP card, having forgotten that what used to be a plastic card is now, increasingly, an app. IBM needs the app so it can get its addled directors the Old Fogey discount on a cup of coffee at one of AARP’s geezer-feeding friends. (Right now Starbucks isn’t on the list of Discounts for the Decrepit.)
If IBM’s management gets smart enough fast enough it will reflect on its history and the history of the computing business as a whole and rediscover what it once knew better than anyone: Sure, there is value in maintaining control of the back end, but the big source of pricing power comes from dominating the gateways. These days the gateways are mobile devices, wearable computers, and local Internet-of-things controllers, and computing devices embedded in cars, appliances, and entertainment gadgets.