The red Porsche's lights are heavily taped and only the most unobservant bystander could fail to notice other racing modifications. Three men huddle around the engine, hands making minute adjustments dictated by sensitive ears. The driver nods satisfaction, listens to last-minute instructions from his crew and slips into the driver's seat. He flexes his gloved fingers on the wheel and adjusts his driving helmet; after years of racing, it still feels constricting. Revving the engine, he checks his tachometer, then slowly pulls the car into the starting area. It's time.
Behind the visor, the driver allows himself a small grin. At age 50, he has a number of very powerful people at IBM wondering just what the hell he's doing. His name is Frank Gerald Soltis. Within the IBM midrange community he is what Michael Jordan was to the Chicago Bulls; he is The Franchise.
Frank was born in Minneapolis. Details of his early years are irrelevant to this narrative; but at the age of 20, while still in the embrace of academia, he was offered a summer job at IBM's Rochester facility. Frank had no interest in computers then, but accepted the job for the same reason countless people accept jobs they don't really want. He needed the money.
It was there he met Harry Tashjian, then-director of the development lab and the management force behind the S/3, S/32, S/34, S/36 and S/38. Tashjian became a mentor to young Frank and convinced him to pursue his doctorate in electrical engineering. Of such ripples are tsunamis formed.
Frank received his Ph.D. from Iowa State and wrote his doctoral dissertation on a revolutionary notion called single-level storage, a concept that would allow programs and data to appear to exist in a single, large main memory independent of distinct technologies like memory or disk.
Duly degreed and dissertated, Dr. Soltis returned to Rochester where he studied the industry research on another revolutionary concept: high-level machine architecture. Under such an implementation, program instructions would be free to operate on objects independent of operands in registers or in storage. Soltis understood that free of the storage's physical properties, instructions could then become hardware-independent, allowing systems to evolve while software remained essentially unchanged.
As genius will, Dr. Frank began the lugubrious process of giving form to his vision. On January 8, 1970, which Soltis remembers as a sunny but Arctic-cold Minnesota day-just three days before the Minnesota Vikings played in their first Super Bowl-he presented IBM management with his proposal for a radically new computer that would combine single-level storage and high-level machine architecture. That proposal became the S/38.
The "technology-independent" architecture of the S/38 allowed for some curious experimentation. Research projects confirmed that it was possible to overlay a S/36 environment on a S/38. The first AS/400 was, in effect, a S/38 capable of running S/36 code. Dr. Frank and his colleagues proposed the next logical step: combining the two systems into a single machine, the AS/400. Steve Schwartz, soon-to-be president of the System Product Division, gave the project his blessing, but it was delayed in the administrative moat that surrounds the castles of innovation at IBM. Finally, in the spring of 1986, with IBM losing midrange market share and with customers impatient for a migration path, Rochester was given the green light-and with it, only 26 months to bridge the chasm between concept and computer.
The biggest challenge was the lack of time. Soltis and other developers worked crushing schedules. During one holiday weekend, management actually had to lock the lab doors and force people to go home and recharge. The innovations required in the new product line were an engineering Everest that could not be scaled alone. Twenty-five hundred technicians in Rochester took up the challenge, with portions of the project farmed out to other IBM installations.
Most critical, however, was IBM's commitment to deliver a product that customers wanted. For the first time in its history, IBM invited customers and business partners to participate in the product-development cycle. Hundreds came to Rochester to share suggestions and requirements; to develop, to test, to ask "What if...?" Pilot systems were installed at customer and business partner sites. What ifs evolved into an integrated operating system. Suggestions were inscribed in microcode. The midnight oil burned in the Rochester lab; but by June of 1988, customers took delivery of the first AS/400s.
I wish I'd had the coffee and Cheetos concession.
According to Dr. Soltis, the AS/400's future was almost its past. John Cocke, the inventor of RISC technology (although it wasn't called RISC at the time), tried putting Soltis' original S/38 on a RISC processor. Limited to a 32-bit implementation, Cocke was unsuccessful and Soltis still regrets not pushing for changes to the RISC processor that would have accommodated the S/38.
But that incompatibility no longer exists. By 1995 the AS/400 will run on a 64- bit PowerPC RISC processor (product of the IBM/ Apple/Motorola alliance), a migration that Dr. Soltis has been spearheading for the past three years. By 1997, both the licensed internal code and OS/400 will be rewritten to be object-oriented. The same 64-bit PowerPC processor will also be used in the high-end RS/6000 and in mainframe Power Parallel processors. Long before the Vikings return to the Super Bowl, Soltis expects to build an AS/400 with the equivalent power of a Cray II supercomputer on a single chip.
In a field where technology be-comes as discardable as yesterday's newspaper, Soltis created a computer architecture that is as viable today as it was 20 years ago. S/38 code, for example, can run totally unchanged on today's AS/400 and will continue to run without modification into the future. The AS/400's high-level machine architecture, single-level storage and object-based programming will allow customers to ride the technology wave without costly modification to their applications.
From the beginning, Soltis ex-plains, his driving motivation has been the protection of the customer's investment and the delivery of a product with long life and adaptability to changing technological trends. In that, he has admirably succeeded and enriched both his corporation and its customers.
The red Porsche slides through the S-turn and accelerates down the straightaway. The man who will take the IBM midrange into the next century is riding the bumper of a Corvette at 150mph. Somewhere, IBM Chairman Louis Gerstner is having a bad case of heartburn and not understanding why. Soltis downshifts, drops underneath the 'Vette, and screams through the turn into the next straightaway. The countryside is a blur of bleeding colors, but Soltis doesn't notice. "Going 150mph requires total concentration," he says. "You can't think about the AS/400."
With special thanks to Dr. Frank Soltis for his time, his candor, and his unfailing good humor; and to Rachel Postlethwaite (spelled just like it sounds) for keeping us honest.
Victor Rozek has 17 years of experience in the data processing industry, including seven years with IBM in Operations Management and Systems Engineering.
LATEST COMMENTS
MC Press Online