Chapter 09 - Networking: Emergence 1979-1981
9.2 Prime Computers
Prime Computers (Prime) initiated a project to create a product interconnecting computers because of a customer request. The project was assigned to Paul Severino, an engineer hired from DEC to design industrial control peripherals to compete with Honeywell, the company the Prime founders had left February 1972.1 Prying Severino from the excitement of DEC took some convincing. Fortunately, Severino wanted to work for a start-up. After designing a number of I/O and data acquisition products, Severino remembers the rush project in 1974:
Prime had a problem. Someone had sold a system that had to have two computers that talked to each other. So what happened was they threw this specification to me and said: ‘OK, we need this done. You know, we’re late. The customer wants this. We don’t have anything to give him. You’ve got to do this project.’ I looked at the spec and it was just a 16 bit parallel interface between two computers. It was clearly done for the computer room. Just interconnect two computers. And I looked at it and I said: ‘Well, why don’t we make this so that it extends a little bit and we’ll make sort of a daisy-chain and we’ll make an eight computer interconnect. I got two computers working and then I jumped on to another project. We shipped it. It was fine. There was no real need to do anything more than two at that time.
Fortunately, William (Bill) Poduska, Vice President of Engineering, instinctively understood the importance of interconnecting computers. Before joining Prime, Poduska had been a professor at MIT involved with Project MAC and familiar with Arpanet. This early exposure to timesharing and computer communications informed his design decisions and proved instrumental in Prime’s eventual success. In late 1974, Poduska gave William Farr, who had also joined Prime in 1972 from Honeywell, some reading material. Farr remembers:
Bill gave me some papers to read to try to generate some ideas, seed my thinking. There were papers by Kleinrock, Abramson, and Kahn and Metcalfe’s PhD thesis. I did a little more literature search and came up with some papers from Bell Labs on SpiderNet, and got really excited about the idea of actually linking computers together in a way that went beyond just serial, RS-232, kinds of communications. We had thought about linking computers with some high-speed parallel cable networks, where there was basically a master and slave. That was very limited, in that, in order to achieve any kind of performance, you had to have a lot of parallelism in terms of the data and fairly high clock rates and expensive cables. The machines also had to be close to each other. Paul Severino had created the Inter-Processor Communication Controller (IPC) that we were selling. At the same time, I was working on these ideas that had developed from the ALOHAnet papers, Metcalfe’s Ethernet papers and the token ring papers.
Farr analyzed the alternative networking schemes and made several presentations to the engineering staff. Concluding in favor of token ring, he wrote an initial product specification. Farr remembers a discussion he had with Poduska in early 1976:
He wanted me to have a network to run at the Spring Joint Computer Conference (SJCC) to be held in June 1976. We had a heated discussion about whether that was a possible or a reasonable thing to do. It could have been done as a basic prototype with little or no software because at that time at Prime things got done that way. You just did it. But since I’m very product oriented, my goal was to get something that would be a robust product, not something that was a prototype. My reasoning at the time was, if we develop this prototype, it would be interesting for the show, but then we’d basically have to go back and start over again and really do something real in order to have a shippable product. Anyway, it didn’t get shown at the SJCC and, as a matter of fact, it took about a year from that point to actually have working hardware.
Poduska challenged Farr not because he wanted to ship an inferior product, but because he knew, or least sensed, the importance of this project for Prime. Farr remembers:
Bill Poduska is one of the most creative people I’ve ever met. He was able to take an idea and push it way beyond what other people were thinking about, and see the value in very practical terms at the same time. He was always pushing. There was never a time when he was satisfied with anything. He would be satisfied with what you had done, but he always realized you could do more if you really wanted to. So there was this constant technological tension that he maintained, but it was a very fatherly type, fostering type. I also worked at DEC in 1961-62 and Ken Olsen was very much the same type of person. As a matter of fact, Ken came around every day and talked to every engineer when I was there. Gordon Bell also was – not quite as social, but very much the same kind of fostering person; very dynamic, generated lots of ideas and encouraged you to pick up on them and continue.
Shortly after the 1976 SJCC, Severino resigned to join another start-up, Prime having grown too big, and the allure of a start-up, this time in the data acquisition business, was too compelling.
But for Farr, Prime remained a very exciting environment. In early 1977, however, his enthusiasm, encountered the resistance of others who did not believe, as he did, in the promise of networking. So despite rumors that DEC, Hewlett Packard and Data General had networking projects under development, he had an all too common experience for those working on leading edge technology. Farr remembers trying to gain the support needed to build a local area network:
I had a boss at that time, Charlie Smith, who said: ‘Bill’s [Poduska] been right often in the past. Maybe I should listen and support this project.’ We went to our marketing organization and they saw absolutely zero need for this product. They thought it was a silly idea. I think the basic reason for that was that nobody else had a network, so how could they sell something that they had no competition for. I said: ‘Can’t you see that this is something that is valuable because people can share information at high speeds?’ They said; ‘Well people don’t need to do that. Nobody’s doing that now, so it’s clearly not important.’ It was very frustrating because until we could get this targeted as a product, we couldn’t get real support to build a large network which was needed to really prove the ideas.
Not to be deterred, Farr convinced Smith to buy enough equipment to build a network to interconnect sixteen computers in the engineering organization. Farr, Bob Gordon and Paul Levine then built a token ring network they called RingNet. By mid-1978, engineering had become totally dependent on RingNet, both for electronic mail and file sharing. In January 1979, Prime announced RingNet as a product.2
- [1]:
The founders had actually been with Computer Control Corporation (CCC), the minicomputer company acquired by Honeywell and the same CCC of Art Carr and John Pugh of Codex.
- [2]:
A year later, Ford Motor Company bought Prime computers for the first time based on the promise of interconnecting a large number of computers.