Chapter 08 - Networking: Diffusion 1972-1979
8.13 National Bureau of Standards and MITRE 1971 - 1979
Whereas computer communications over the public telephone networks preoccupied the Europeans, the American standards making body, the National Bureau of Standards (NBS), focused on the emerging technology of private local area networks. This was not the result of some bureaucratic deliberation but the initiative of largely one man: Robert Rosenthal.
After the NBS installed their Arpanet TIP in late 1971, Rosenthal was assigned the responsibility for building instrumentation to measure network performance: not of the network itself, which was being done by the NMC at UCLA, but performance from the perspective of the user. NBS wanted to understand the value of networking in order to recommend its use to other agencies, a responsibility consistent with the NBS charter of:
developing federal information processing standards, assisting other agencies deploy technologies that support those standards and conducting appropriate research to assure federal agencies remain on the leading edge in the use of technology.
By 1976, Rosenthal, familiar with the work going on at PARC, began thinking of how he might build local area networks suitable for use in the Federal government.58 With full support from his superior Dr. Ira Cotton (who, as Rosenthal fondly remembers, said: “Here’s some rope, go hang yourself!”), Rosenthal contacted Metcalfe, Boggs and Shoch at PARC as well as others knowledgeable of networking, such as Charlie Bass then at Zilog. Rosenthal remembers:
Our concept was to take really dumb terminals, give them to employees all over the Bureau of Standards, and let them access the few very large hosts that we had available.
By 1978, three working prototype boards had been built and a Request for Proposal (RFP) to build production quantities was let and won by a small company in Florida. In late-1978, as the network boards began being received from the vendor and installed, NBS-Net came into being. Rosenthal comments:
So we implemented a network based loosely on the technology of Ethernet. It had the unique attribute of having a single power supply, fanning out to up to eight RS-232 connectors, which at the time was completely different from the Xerox approach of building an Alto like workstation supported by a local network technology.
Rosenthal recalls:
I got so excited about all of this that I thought it would be a good idea to put a workshop together. So I called everybody I knew in local area networking. MITRE was one of the first calls.
MITRE, of Bedford, MA, a government contractor for primarily the Air Force, had installed its Arpanet TIP at the same time as NBS, at the end of 1971. Concurrently, MITRE received one of two contracts let by the Air Force to develop local area networking prototypes and conduct networking studies. The other contract went to Ford Aerospace, of Sunnyvale, CA. By year-end 1972, Ash Dohad of MITRE had working a slotted broadband network named MITRE-Net. MITRE-Net used radio frequency (RF) technology over a cable to create many side-by-side communication channels much like frequency division multiplexing. Broadband differed radically from Ethernet and token ring in that it multiplexed many channels onto the transmission media. Ethernet and token ring created just one communication channel over the coaxial cable or twisted-pair wire.
Then in 1976, Greg Hopkins engineered changes to MITRE-Net so one or more channels could service Ethernet-type traffic. Although the performance of the Ethernet-like channels was not impressive - three hundred thousand bits per second - MITRE did secure a patent for its contribution. (Their work cites the prior work of Abramson and Metcalfe.)
In late-1978, the Air Force requested MITRE investigate the issues of local networking and recommend actions they, the Air Force, should take. The timing proved propitious, for shortly thereafter came the call from Rosenthal suggesting they hold a workshop. It did not take long to agree to convene a forum on “Local Network Protocols” to be held January 31, 1979 in Columbia, MD. In their subsequent calls to other possible participants, it became very clear that while virtually everyone who had come into contact with local area networking sensed its importance, they all expressed confusion over what it all meant or what to do about it. Perfect fodder for a forum.
- [58]:
Rosenthal: “From my perspective, it was: ‘Hey, I’ve got a group here and I’ve got to get real smart about this technology.’ We had a number of contracts with some other agencies to actually install some of the early three megabit Ethernet devices, and I had put a lab together here with Altos and Dover printers and the like. We also did some work for some people downtown. So we were very much aware of what Xerox was up to. Xerox’s mindset, as I recall, at the time, was not to unbundle the LAN technology, but to sell an office system. Our motivation was to unbundle that technology and provide the equivalent of a carrier, but local area networks. At the time – we were forcing definitions like ‘locally owned and administered.’ All the bad things that we knew about carriers, from a user’s perspective – they’re not bad; the regulatory kinds of things – we wanted to do without, because we were trying to connect terminal devices within buildings we owned. There was no need for carrier services, so we administered them ourselves, we did everything ourselves. That’s what we meant by local networks at the time.”