ENDPGM > 2008

A Triumph of Hard Work

Frank Soltis Reflects on the 20-Year Anniversary of the AS/400

Even IBM* Chief Scientist Frank Soltis, the father of the AS/400* server, admits he never expected to be discussing the business computer a full two decades after its unveiling.

“To think then that I’d be sitting here, talking about the 20th anniversary? It’s like, ‘Wow, that would never happen,’ ” Soltis says. “I don’t think any of us looked ahead.”

When he looks back on those hectic months leading up to the server’s introduction on June 21, 1988, Soltis still marvels at what the technologists at IBM Rochester accomplished. They fought for backing for a radical idea, they fought for their jobs and, along the way, they often fought one another. Ultimately though, the technology they advanced continues to run businesses worldwide.

As long-standing midrange professionals know, the business computer called the IBM AS/400 server evolved from two IBM products, the System/36* and the System/38* servers. If you think consolidating those two systems was challenging, imagine combining five. That was IBM’s plan in the early 1980s. In the interest of efficiency, the company wanted to combine all of its computing platforms—including System/36, System/38 and the mainframe—into a single system code named Fort Knox.

“It was one of those overly ambitious projects that just wasn’t going to work,” Soltis says. “You’d look and you’d say, ‘Wait a minute, I can’t satisfy the needs of all five of these systems’ customers with one system.’ ”

Soltis and other skeptics—to name just two, Dick Mustain, a top System/36 architect, and Dick Bains, an expert in compiler technology and languages—were taken off Fort Knox. Symbolically, their new advanced-technology group was housed not at the main IBM Rochester plant, but across the road in the then-new “white buildings.”

“You could tick off the various people who were instrumental in both the System/38 and the System/36 servers, and they ended up in this advanced technology group,” Soltis says. “Every one of us believed Fort Knox wouldn’t work.”

Eventually, IBM reached the same conclusion. By then, the advanced-technology group was nurturing the idea of running System/36 applications in a System/38 environment.

Technologists from the main IBM plant would trek over to the white buildings before or after their regular shift to lend their expertise. They weren’t motivated just by the pursuit of new technology. They also believed jobs were at stake. That’s because at the IBM New York headquarters, Rochester’s future was tied to Fort Knox.

“When Fort Knox collapsed, there was nothing here,” Soltis says. “That would have been the end of Rochester products.”

Convinced that they were saving IBM Rochester itself, employees worked tirelessly on what, in late 1985, became the Silverlake project. Compressing a typical five-year development cycle into barely two years, IBMers practically had to be dragged away from their work.

Soltis recalls IBM locking the plant for an extra day over one holiday weekend just to keep people out. “People were putting in an incredible number of hours and not necessarily being paid for it,” he says. “But the thing was, we used to do annual opinion surveys, and at that time we had the highest opinions ever, as far as people being happy. We were content with what we were doing.”

Content, even amid contentious moments among developers of the two systems. Even though multiprocessor design and I/O architecture came from the System/36, the AS/400 was, at its core, an enhanced System/38. And many System/36 developers doubted this direction.

But Soltis describes the whole experience as IBM Rochester showing what it could do. That’s one reason he found such satisfaction in seeing the first AS/400 systems roll off the assembly line.

“When we actually announced and began to ship the AS/400 server, there was a tremendous feeling of, ‘This is ours. We did this,’ ” he says. “To me it was an exciting time.”