Back to top

Chapter 09 - Networking: Emergence 1979-1981

9.4 Robert Metcalfe and the MIT Laboratory of Computer Science

Metcalfe soon found his first consulting client, Michael Dertouzos, of MIT’s Laboratory of Computer Science (LCS). Dertouzos faced some local area networking decisions in as complex a computer environment as existed at the time. As a result of a university gift program begun in 1978 by Xerox, LCS (along with computer science laboratories at Carnegie-Mellon and Stanford) received Altos workstations, file servers and the experimental 3 Megabit per second Ethernet. That meant that LCS had two networks, Ethernet and token ring, with another network, Chaosnet, at the nearby AI Labs. DARPA, keenly interested in the comparative advantages of the different networks, awarded LCS a contract to conduct performance measurements. In addition, DEC proposed to give LCS twenty to thirty DEC VAX 11/750 minicomputers if used as single user workstations. To make the VAX’s effective, however, required interconnecting and integrating them into their existing computer facilities. The most obvious, but not necessarily the best, decision seemed to be to reengineer a faster, second-generation token ring; a technology unencumbered by the complications of Xerox’s uncertain commercialization policy.

Metcalfe helped Dertouzos, not as an engineer, but by evaluating management practices and recommending changes. He would advocate that LCS agree to DEC’s terms and commit to a vision of desktop computing, even if it meant interconnecting the VAX’s with token ring and not Ethernet.

Interconnecting the VAX’s with token ring required upgrading the performance of their existing 1 Megabit LNI’s. To do so meant finding outside help, for Dertouzos knew LCS did not have the requisite skills to do the re-engineering in-house. For help, Dertouzos thought of Howard Salwen, a former classmate, who had a small company that engineered communication products principally for government contracts. Dave Clark recalls:

Mike knew that Howard’s company was very involved in communications and, in particular, that Howard was very strong in the analog engineering skill, and he knew that the crucial aspect of getting this ring to work right was to have somebody who is not just a digital engineer. Mike was very anxious to make sure that we succeeded by finding somebody who would match our skills. He basically said: ‘I know Howard. I’ve known him for a long time. He’s a good guy. He run’s this consulting company. We’ll give him a contract for $100,000 to do a Version 2 LNI for the Unibus.’ And so we wrote him a contract. We didn’t do any competitive selection or anything like that.

Salwen and a partner had founded Proteon Associates (Proteon) in 1972. In 1974, health problems forced his first partner to resign and Al Marshall became Salwen’s new partner. Proteon’s competence rested largely in designing and building modems under a number of government agency contracts. Salwen remembers his early involvement with MIT:

There were meetings at MIT that were attended by Metcalfe, who at that time was consulting to LCS. The meetings were to discuss a philosophy that would be adopted by the whole group for networking. Of course, Metcalfe was strongly in favor of Ethernet. There was a group from the Artificial Intelligence Lab arguing in favor of their Chaosnet design. Dave Clark and Jerry Saltzer were in favor of token passing ring. I came in as a consultant with Al and started attending these meetings. And the meetings were very exciting. There was a lot of screaming and yelling in those meetings.

Metcalfe’s consulting to a university, especially one favored by Xerox, may have raised eyebrows at Xerox, but the possibility of his working for a competitor raised fearful concerns. Those prone to yell fire at the first sign of smoke must have ranted “I told you so” after reading Metcalfe’s newsletter promoting his availability and expertise, and advocating the coming of networking and the superiority of Ethernet. Given Xerox’s proprietary attitude regarding their technology, it would seem a challenge only a masochist would have undertaken. But Metcalfe had two advantages: his own unrelenting drive to materialize his vision, and an ally and friend in David Liddle.

As Liddle recalls:

Bob wanted to somehow stimulate things happening out there. That was OK with me provided Xerox got something out of it. I didn’t want him to just go do it to make a big, blooming consulting business and have Xerox not benefit from it. So he sort of wanted to be the marriage broker of getting Xerox to open up the product to lots of other customers, lots of other companies. My position was: ‘No, but a few selected companies that could make really strong partners so we can make people see this as a de facto standard, then that’s fine.