Aqua
04-07-2004, 11:53 AM
IBM Marks 40th Anniversary Of Biggest Bet
By Ken Spencer Brown
Bob Evans knew something big was brewing that chilly Milwaukee day in December 1960. A senior-level IBM exec had told him to drop everything and fly to New York for an 8 p.m. meeting.
Evans didn't know the call would lead to one of the biggest gambles in IBM's (NYSE:IBM - News) history - maybe one of the biggest ever by any tech firm.
The result: a breakthrough family of systems that would secure Big Blue's place in the fledgling computer industry and spawn a whole new segment.
Long before Linux (news - web sites), Unix (news - web sites) and Windows battled for corporate dominance, before the dot-com boom and bust, before computers became small enough to fit within a cubicle - before businesses even had cubicles - there was the IBM System/360.
The first general-purpose mainframe, which turns 40 today, cost IBM nearly $30 billion in today's money to develop and launch. IBM hired 60,000 people and opened five new plants.
The bet paid off handsomely. Within two years, IBM had 1,000 orders per month. And years after rivals have gone under or gotten out of the business, IBM continues to rake in billions from mainframes. IBM sold $4.2 billion's worth during 2003, says International Data Corp.
The machine helped send man to the moon, modernize air travel and pave the way for the personal computer.
And it almost didn't happen at all.
Waiting for Evans in New York was T. Vincent Learson, a 6-foot-6 Harvard-trained math whiz who would go on to head IBM a decade later. Learson asked for a progress report on a different machine - the company's planned next-generation system.
Engineers, including Evans' boss, had their hearts set on what was to be the 8000 series. The firm had already built one model and was just months away from a formal launch.
But Evans didn't like what he saw. For one, he wanted a family of computers that could use the same hardware and software. The 8000 couldn't do that.
The idea of compatibility was radical at the time. Critics said computers couldn't be truly great if they used lowest-common-denominator technology.
But Evans had fought to make IBM's 1401 and 1410 compatible a few years earlier, and customers had embraced the idea.
Some rivals understood this better than IBM itself. Honeywell's low-end H-200 could run IBM 1401 code with a little tweaking, thanks to Honeywell's Liberator software.
The new approach helped IBM cut costs. By creating variations on a common platform, Big Blue could address different segments of the market without duplicating engineering efforts.
"We spent 90% of our money in 1960 just evolving the dissimilar (processor) families," Evans said. "About 10% was left for disks and tapes and almost nothing for programming."
A common architecture freed much of that remaining 90%. By 1980, IBM's spending was about evenly spread among mainframes, peripherals and programming.
The shift also made the 360 the first general-purpose mainframe. Earlier computer families were designed for specific segments, such as science research or accounting.
That changed the role of software. Now programmers could write code without having to worry about what model of hardware it would run on.
More important, it sparked new demand for computers. Businesses that couldn't afford to rent a top-of-the-line system could start small with a cheaper, low-end 360. If they needed a bigger machine later, they could trade up without ditching all their old software and hardware.
"It was the beginning of companies using computers as a resource for maintaining business data," said IDC analyst Steve Josselyn.
Still, it was a risky move. Any problems in the 360 would affect a range of IBM's segments, not just one set of customers. Keeping machines compatible also locked IBM into early design choices; bad ones could come back to haunt the firm years later.
That appeared to be the case for the first few months after IBM announced the 360. Few were selling. Evans worried that the firm had produced something too advanced for its time.
He breathed a little easier by July, when orders began picking up. The trickle of orders became a flood.
"It sold beyond our wildest expectations," Evans said.
Today, mainframes are still a big business for IBM.
Aside from the money it brings in, the mainframe unit develops technology that later makes its way into cheaper systems.
"It is the DNA of IBM," said Erich Clementi, general manager of the firm's zSeries, the latest descendant of the 360.
The 360 was the first mass-produced system to use solid logic technology, a precursor to the modern computer chip. And the model 40, part of the 360 family, was the first to boot from a floppy disk.
Concepts such as clustering, workload balancing, virtualization and grid computing are touted as big advances for Linux systems. They're a given in the mainframe world.
Even as personal computers and midrange Unix systems get all the attention, mainframes live on, Clementi says. When a customer needs an industrial-strength system that doesn't break down, clustered PCs and even Unix servers often don't cut it. Meanwhile, mainframes have changed with the times. The xSeries can run Linux, for example, and fits well in a grid setting with other systems.
"The mainframe market wouldn't be alive if it were still what it was in the '60s," Clementi said.
That sort of flexibility, mixed with mainframe muscle, gives it staying power.
By Ken Spencer Brown
Bob Evans knew something big was brewing that chilly Milwaukee day in December 1960. A senior-level IBM exec had told him to drop everything and fly to New York for an 8 p.m. meeting.
Evans didn't know the call would lead to one of the biggest gambles in IBM's (NYSE:IBM - News) history - maybe one of the biggest ever by any tech firm.
The result: a breakthrough family of systems that would secure Big Blue's place in the fledgling computer industry and spawn a whole new segment.
Long before Linux (news - web sites), Unix (news - web sites) and Windows battled for corporate dominance, before the dot-com boom and bust, before computers became small enough to fit within a cubicle - before businesses even had cubicles - there was the IBM System/360.
The first general-purpose mainframe, which turns 40 today, cost IBM nearly $30 billion in today's money to develop and launch. IBM hired 60,000 people and opened five new plants.
The bet paid off handsomely. Within two years, IBM had 1,000 orders per month. And years after rivals have gone under or gotten out of the business, IBM continues to rake in billions from mainframes. IBM sold $4.2 billion's worth during 2003, says International Data Corp.
The machine helped send man to the moon, modernize air travel and pave the way for the personal computer.
And it almost didn't happen at all.
Waiting for Evans in New York was T. Vincent Learson, a 6-foot-6 Harvard-trained math whiz who would go on to head IBM a decade later. Learson asked for a progress report on a different machine - the company's planned next-generation system.
Engineers, including Evans' boss, had their hearts set on what was to be the 8000 series. The firm had already built one model and was just months away from a formal launch.
But Evans didn't like what he saw. For one, he wanted a family of computers that could use the same hardware and software. The 8000 couldn't do that.
The idea of compatibility was radical at the time. Critics said computers couldn't be truly great if they used lowest-common-denominator technology.
But Evans had fought to make IBM's 1401 and 1410 compatible a few years earlier, and customers had embraced the idea.
Some rivals understood this better than IBM itself. Honeywell's low-end H-200 could run IBM 1401 code with a little tweaking, thanks to Honeywell's Liberator software.
The new approach helped IBM cut costs. By creating variations on a common platform, Big Blue could address different segments of the market without duplicating engineering efforts.
"We spent 90% of our money in 1960 just evolving the dissimilar (processor) families," Evans said. "About 10% was left for disks and tapes and almost nothing for programming."
A common architecture freed much of that remaining 90%. By 1980, IBM's spending was about evenly spread among mainframes, peripherals and programming.
The shift also made the 360 the first general-purpose mainframe. Earlier computer families were designed for specific segments, such as science research or accounting.
That changed the role of software. Now programmers could write code without having to worry about what model of hardware it would run on.
More important, it sparked new demand for computers. Businesses that couldn't afford to rent a top-of-the-line system could start small with a cheaper, low-end 360. If they needed a bigger machine later, they could trade up without ditching all their old software and hardware.
"It was the beginning of companies using computers as a resource for maintaining business data," said IDC analyst Steve Josselyn.
Still, it was a risky move. Any problems in the 360 would affect a range of IBM's segments, not just one set of customers. Keeping machines compatible also locked IBM into early design choices; bad ones could come back to haunt the firm years later.
That appeared to be the case for the first few months after IBM announced the 360. Few were selling. Evans worried that the firm had produced something too advanced for its time.
He breathed a little easier by July, when orders began picking up. The trickle of orders became a flood.
"It sold beyond our wildest expectations," Evans said.
Today, mainframes are still a big business for IBM.
Aside from the money it brings in, the mainframe unit develops technology that later makes its way into cheaper systems.
"It is the DNA of IBM," said Erich Clementi, general manager of the firm's zSeries, the latest descendant of the 360.
The 360 was the first mass-produced system to use solid logic technology, a precursor to the modern computer chip. And the model 40, part of the 360 family, was the first to boot from a floppy disk.
Concepts such as clustering, workload balancing, virtualization and grid computing are touted as big advances for Linux systems. They're a given in the mainframe world.
Even as personal computers and midrange Unix systems get all the attention, mainframes live on, Clementi says. When a customer needs an industrial-strength system that doesn't break down, clustered PCs and even Unix servers often don't cut it. Meanwhile, mainframes have changed with the times. The xSeries can run Linux, for example, and fits well in a grid setting with other systems.
"The mainframe market wouldn't be alive if it were still what it was in the '60s," Clementi said.
That sort of flexibility, mixed with mainframe muscle, gives it staying power.