Chapter 2 - Background
2.22 The IBM System/360 and the Third Generation of Computing --1964
On April 7, 1964, Thomas Watson Jr. and the management of IBM made the most important product announcement in their company’s history.112 IBM would begin shipping six models of a revolutionary new computer architecture named the System/360 in April 1965. At a cost to IBM of an estimated $5 billion, the System/360 represented the transition to the Third Generation of mainframe computers, traumatized IBM’s customers, and sent IBM’s competitors scrambling for survival. The effect was so cataclysmic as to preclude IBM from ever again introducing a new computer system that was incompatible with its existing computers.113
In 1960, IBM’s internal organizational structure mirrored the major segmentation in the market for computers: beginning computers and large general purpose computers. IBM’s two computer divisions – the General Products Division (GPD) selling beginners’ computers, and the Data Systems Division (DSD) selling general purpose scientific and business computers – reflected external market segmentation because the two had co-evolved. But IBM’s organizational structure had become a source of serious problems.
GPD sat pretty. In 1960, the 1401 was the”most popular” IBM computer according to Fortune Magazine, 1966,114 and division management felt they would not need an improved 1401 until 1968 – a strong indication that management feared no threat of innovation from others.
The same could not be said for DSD. DSD management knew they had to respond to competitors’ technological improvements, and soon. The 7000 series of computers needed to be upgraded – preferably to a new generation of products. The question was to what?
In March 1961, DSD executives sought authority to begin developing the 8000 series – designed to succeed the 7000 series. But IBM’s corporate management committee said no. To some key members of the committee, the 8000 represented more continued product proliferation. And even though IBM had achieved market leadership by, in part, a proliferation of products, it had to end. The proliferation of models, each of which was to some degree a unique development effort, had become unmanageable. Furthermore, GPD and DSD increasingly competed with each other. The concerned executives wanted most of all to sell one computer architecture to the full spectrum of customer requirements, all using the same peripherals and software. They argued the moment called for “drastic simplification.”115
Corporate management first turned to their Advanced Systems Development Division – the division responsible for innovating technologies for future products. Their recommendation to use cryogenic technologies was rejected as well.116
By August 1961, Watson Jr. and Learson, CEO and COO respectively, were concerned because an alternative to the 8000 had not been created. They started an investigation. (Their anxiety, no doubt, increased knowing it was the same management team that had bungled the STRETCH project.) Their suspicions that no new architecture was being created were confirmed. In November, they convened a task force of thirteen senior officers to recommend action. On the fourth of January, 1962, the task force gave their recommendations to Watson Jr. and Learson. As reported by Fortune Magazine a few years later (1966):
- There was a definite need for a single, compatible family of computers.
- The new line should not be aimed at simply at replacing the popular 1401 or 7000 series, but at opening up whole new fields of computer applications.
- The System/360 must have both business and scientific applications. Information input and output equipment, and all other peripheral equipment, must have a “standard interface.117
The report estimated $125 million in programming costs at a time when IBM spent $10 million a year in programming. After hearing the recommendations and discussion, Learson concluded: “All right, we’ll do it.”118
The simple declarative – “All right, we’ll do it.” – the one that cost $5 billion – dramatically changed IBM. According to Fortune Magazine, the shift “plunged I.B.M. into an organizational upheaval.”119 IBM, forced in part by competitive pressure, yet also incentivized by confidence in the economic potential latent in innovating computers, transformed itself into a new company. In tackling the System/360 project, IBM used a development project to recast itself, and the economic and technical world in which it competed. The article goes on to list some I.B.M. changes:
- The company, which was essentially an assembler of computer components and a business-service organization, has now also become a major manufacturing concern as well.
- The view now is that I.B.M. is a fully integrated international company, in which the managers of overseas units are presumed to have the same capabilities and responsibilities as those in the U.S.
- Now a new group of technically oriented executives has come to the forefront for the first time, diluting some of the traditional power of marketing men in the corporation.
- I.B.M.’s balance sheet has a new look.120
By investing in large-scale production, distribution channels, and management structures, IBM had secured “first-mover” advantages.121
Innovating the System/360 required a new computer architecture(s) as well as the mastery of important new component technologies – performance objectives could not be met using then current component technologies. Previous work on the STRETCH proved invaluable.122 Innovations included:
- Not just an operating system, but a common one across all models.123
- Six models all to be introduced at once. The internal decision making and organization coordination were unprecedented.
- Able to process both business and scientific applications.
- Common Input/Output interfaces across all models.
- Standardization on an eight-bit byte architecture, and
- First commercial use of MIMC technologies.
The last three came directly from STRETCH. To be discussed will be MIMC. The importance of the eight-bit byte will resurface later.
The System/360 had to be built with post-transistor component integration if the performance objectives were to be achieved. IBM was fully aware of the developments in integrated circuits. It also had its own experience building hybrid integrated circuitry for STRETCH called MIMC. In MIMC, transistors and diodes were still made separately. Then they were soldered into place into modules, which were then assembled together on printed circuit boards. The task force assigned to recommend use of either integrated circuits or MIMC decided IBM could not scale manufacturing of integrated circuits rapidly enough to meet production schedules, and recommended the less risky MIMC.
Before the System/360 and MIMC, IBM bought standard components from manufacturers such as Texas Instruments. IBM then assembled the components, by wiring them together, to construct the computer architecture and create a computer. With MIMC, IBM had to make a decision. If they contracted with independent manufacturers to construct the MIMC modules, they were giving proprietary design knowledge away. The alternative was to manufacture the MIMCs themselves. “Unless we did it ourselves, we could be turning over some of the essentials of our business to another company,” said Watson Jr. in Fortune Magazine.”We had no intention of doing that.” I.B.M. also then captured the profits of manufacturing, a significant amount if the System/360 was to be as successful as they thought.124
With a single computer architecture – made possible by one, common operating system – it no longer made sense to be organized into a General Products and Data Systems divisions. GPD and DSD were combined into the Data Processing Group, which, according to a 1968 article in Business Week, accounted for “ 79% of IBM’s sales and at least 85% of earnings.”125 The DP group was further organized into four regions, with specialized units in thirteen industry sectors and one unit for each major type of 360.126 IBM would be competing with one voice, coordinated by territory and supported by industry sector and 360 type personnel.127 The scale and scope of attention IBM was able to bring to an average sale would dwarf whatever any competitor could do. As evidence of IBM’s dominant market-structure position, in 1969, IBM was shipping “over 1,000 model 360 systems a month.”128
The System/360 had a profound affect on both IBM’s customers and competitors.
For IBM’s customers, the System/360 was a nightmare. MIS managers had to sell a sizable new capital expense, fight for increased operating expenses, re-write essentially all of their application software, integrate new peripherals into operations, hire and train personnel, and manage a significantly more complicated and complex system and operations. Combined, these phenomena became known as: “reset to zero.”129
From the perspective of time, an MIS history of this period went: “In 1969, we delivered what we promised in 1967 for use in 1968. In 1964, we ordered third-generation equipment, in 1965-66 we installed it. In 1966 and 1967 we were totally devoted to converting and learning (in that order).”130 The problem under-estimated by IBM was software. A satisfactory, and stable, release of the operating system took until 1968 to finish. In the intervening years, the beleaguered MIS departments had to either make their old, now obsolete, systems suffice, or run their software in emulation mode on a System/360 – a painfully slow emulation mode it should be added. In a March 1966 press conference, Watson Jr. conceded that “the momentous decision to announce the entire 360 package at once in April 1964, may have been ‘ill advised’“131
The competition could do little at first other than wage a war of words. They argued that IBM could never deliver, it was too expensive, and it was not even state-of-the-art – it didn’t use integrated circuits, for example.132 But all were forced to develop Third Generation product lines to stay competitive. One way competitors tried to differentiate their products from System/360 was timesharing – largely because IBM did not support timesharing in the announced System/360s. That story and other competitive responses comes next. It should become clear that IBM not only transformed itself with the System/360, it transformed the market-structure of mainframe computers as well.
“The Rocky Road to the Marketplace,” Fortune October 1966, p. 138
The IBM System/360 thus qualifies as a technological discontinuity.
“I.B.M.’s $5,000,000,000 Gamble,” Fortune September 1966, p.118
Ibid., p. 123
Ibid., p. 139
Ibid., p. 228
Ibid., p. 228
Ibid, p. 143 O
Ibid, September, p.121 A new look to the Balance Sheet was the existence of debt. TBD
A concept introduced by Alfred Chandler, a preeminent business historian. See “The Visible Hand,” and “Scale and Scope.”
Fortune Oct 1966, p.212
Users had begun innovating operating systems much earlier, with the IBM 704. What was new by way of scale was a manufacturer innovating, selling and supporting an operating system.
Ibid, p. 139 and 142 O
Business Week, June 15, 1968, p. 89
By 1967, IBM’s new organizational structure and the design of its corporate information system reinforced each other; each having changed to accommodate the needs of the System/360.
Business Week, June 15, 1968, p. 89
John Diebold, “Bad decisions on computer use,” Harvard Business Review Jan/Feb 1969, p.14
Philio A. Dorn, “The Onslaught of the Next Generation,” Datamation Jan 1970, p. 76
Datamation, 1969: An overview of the news,” January 1970, p.89
Fortune Oct 1966, p. 211
Ibid., p. 206