Dan Warmenhoven earned a BS in electrical engineering from Princeton in 1972 (in the same class as David Boggs, co-creator of Ethernet). Warmenhoven started at IBM immediately after graduation, working on distributed systems and communications systems for projects like the IBM Series I and the IBM 3790. In 1977 he moved to the Communications Products Division in Raleigh, NC, where he worked on the remote job entry (RJE) package for the IBM 8100, the host computer for IBM’s distributed processing system.
In ’79, Warmenhoven was transferred to the newly established Network Products division, under development manager Murray Bolt. This new division was tasked with developing system independent communications products. Warmenhoven’s first responsibility was engineering manager of the IBM 3710 Network Controller. In 1980, following the announcement of the Ethernet standard by DIX (the consortium of DEC, Intel, and Xerox), IBM established a task force to formulate a response to this innovation in local area networking. Warmenhoven was part of this task force, and in mid ’81 he became manager of the System Design and Architecture division, in charge of managing the implementation of IBM’s token ring alternative to Ethernet across multiple IBM products.
Warmenhoven describes the decision of IBM’s senior management to select an alternative to Ethernet for its networking solution, remembering the mindset of management regarding Ethernet and the DIX consortium: “It wouldn’t be so bad if it was Xerox only, but having DEC in the fray, that was like a declaration of war. I mean, the Axis Powers had formed, and so it had to be, IBM had to have a different solution.” He recounts important decisions in the development of the IBM token ring product, including the need to integrate with IBM’s existing communication technology, SNA (System Network Architecture); how to solve the problem of multiple existing cabling systems; and the requirement of token ring to be available for non-IBM device attachment. This last requirement was anathema to the company’s long-standing practice of offering only proprietary technology. It also led to IBM signing an agreement with Texas Instruments in December 1982 to produce a token ring chipset for the open market, and to propose a token ring standard to the IEEE 802 committee responsible for networking standards.
I spoke with Dan at my home office in Atherton. He had left IBM in ’83 and was then working at HP. In the interview, he spoke frankly and without any apparent bias toward his former employer. This interview provides a valuable perspective on the response of IBM, the major incumbent of the computer industry, to the disruptive technological shift brought on by the entrepreneurs of local area networking.
Keywords: IBM, token ring, 3710 Network Controller, System Network Architecture (SNA)