An Enabling Institution of Local Area Networking
9.10 ISO/OSI (Open Systems Interconnection): 1982 - 1983
In the summer of 1982, Graube and Rosenthal began focusing on what they saw as the next challenge in local area networking: the higher layer protocols. The two shared concerns over the timeliness of higher-layer software, especially the transport layer, the layer required for reliable end-to-end communications over a network. Graube remembers reflecting after it had become clear that at least one LAN standard, CSMA/CD, would be passed:
‘’What do we do for these higher-layer protocols?’ Obviously, it was just like with the 488 standard business, where we had the transport mechanism, but the instruments still couldn't communicate because there was no data format specified. So I wanted to see what we could do in terms of getting some agreements, what have you, for these higher-layer protocols; what was really needed.”
Rosenthal also recognized the need for testing of the protocols that each vendor would create. An OSI standard, while necessary, was not sufficient condition to assure commercial LAN products. Standards specifications could be implemented in an unlimited number of ways, hence the near certainty of incompatible implementations. To solve that problem, Rosenthal saw the use of agreements -- contracts -- among organizations to implement compatible-versions of transport protocol standards. But how would he get organizations to agree? Rosenthal remembers:
“In order to pull this off, we had to get some agreements. That's the key word, 'agreements.' We had to get the people highest up in these organizations to commit resources. We had to get a commitment of the CEOs, somebody with signature authority, had to be able to say: "Here's the check, you make it happen. Pull out all the stops. OSI is important. Make it happen." We had to get the technical people to ask the question: "Make what happen?" We had to say: "Make this happen," and we had to lay it out for them.”
One CEO who responded to the invitation of NBS to attend a meeting was Roger Smith of General Motors (GM). GM had formed a task group in 1980 to look at the possibility of using computer-based automation to combat growing competition from companies producing in lower labor cost economies. In 1981, GM held exploratory conversations with IBM, DEC and HP. From these efforts came the release of GM’s Manufacturing Automation Protocol (MAP) version 1.0 in 1982. Since MAP operated over the LANs that interconnected automation equipment, GM was eager to hear what NBS had to say.
An ideal opportunity to discuss these issues with a wider community was after the IEEE 802 meeting held at DEC in December 1982, Graube convened a handful of concerned insiders to discuss how to create OSI transport layer software. Representatives of roughly six companies, Rosenthal and Graube attended. While all expressed concern, no one knew precisely what to do and little came of the meeting until shortly after Graube submitted his trip report at Tektronix. Graube remembers:
“The lightening bolt came down from the lawyers, through my boss, to me, and "Zap!. Though shalt not do things like that anymore!"“
Tektronix lawyers had recently dealt with the Department of Commerce on another standards matter and issues of antitrust had surfaced. Questions of standards preventing trade and of monopolistic practices had cooled Tektronix lawyers to the idea of companies meeting, clandestine or not, to discuss and create standards.
Graube remembers telling Rosenthal:
"‘Gee, I got my fingers slapped here for our little meeting in Tewksbury,’ and Robbie says: "Well, no problem. We hold these workshops at NBS all the time on various topics. Why don't we hold an NBS workshop on implementers of local area networks?" That's where it started, basically. It works nicely because the National Bureau of Standards works under the Department of Commerce, where the anti-trust division is also located.”
In February 1983, the first NBS International Workshop for Implementers of OSI met. Graube, viewed as the protagonist, was elected Chairman. Conversation quickly turned to the difficulties of ensuring compatible implementations and the desirability of motivating cooperation. Then similar to Roberts and Kahn’s motivation in 1972 to stage a public demonstration of Arpanet at the ICCC trade show, those attending the NBS OSI workshops came to the same conclusion: a public demonstration to coalesce cooperative behavior. They knew it would take time to organize, and settled on the largest computer tradeshow, the National Computer Conference (NCC) to be held in July 1984.
The collision of LAN interests and antitrust concerns could be read elsewhere when in February, IBM offered to buy 12 percent of Intel for $250 million and the Justice Department and Federal Trade Commission made known their reservations. IBM publicly stated its investment was to assure that Intel, a leading U.S. semiconductor manufacturer and supplier of the microprocessor used in the IBM PC, had the capital to stay competitive with the aggressive Japanese semiconductor manufacturers. (IBM accounted for 13% of Intel’s sales.) But could IBM have been hedging its LAN bets?
In May 1983, ISO formally approved the Reference Model standard: ISO 7498. Already in process was an Addendum to the Reference Model to cover connection-less transmission. SC6, which had focused on creating a connection-oriented network layer protocol (8348) to submit as a DP, began work on a connection-less Addendum to 8348.
As of the summer of 1983, ISO had standardized the Reference Model and approved a connection-less transport protocol. They had yet to decide what to do about the LAN protocols being developed within IEEE, and supported by ECMA, or a connection-less network layer protocol. Nevertheless, the promise of standards bringing an end to LAN confusion seemed almost tangible.