Chapter 11 - Standards: An Enabling Institution 1979-1984
11.11 The Emergence of Technological Order: 1983 - 1984
The years 1983-1984 represented a turning point in the history of LANs. The technological-economic dance of chaos and uncertainty shifted into standards resolution and economic growth. The first to benefit were Ethernet vendors. In 1983, Ethernet (CSMA/CD) became an IEEE, ECMA, and effectively an ISO/OSI standard. In addition, the conversion of Arpanet to TCP/IP on January 1, 1983 represented a milestone for DARPA and that TCP/IP had been successfully ported to all the leading computers of the day. By mid-year it would be available for the IBM PC. In contrast, Xerox refused to release more of XNS and, as a consequence, nearly all the LAN start-up’s would engineering their next generation products using TCP/IP not XNS. As for IBM, the lumbering giant, it would not be until 1984 before they make their intentions clear. That same year, the NCC public demonstration of OSI software was an important first step in proving the concept of vendor-independent OSI LAN software; albeit commercial products were still years away. As the finalization of standards became apparent, sales of LAN products soared 141% in 1983. Two years later they reached nearly $1 billion. Here then is a summary of those two critical years.
On June 23, 1983, a day many had fought long and hard for, the IEEE ended years of debate and misdirection and approved CSMA/CD as standard 802.3. DIX and the corporate support Ethernet had garnered prevailed. It was the news supporters within ECMA had sought. With renewed energy they pressed their arguments that ISO adopt the CSMA/CD standard. The fact was: Ethernet worked despite years of arguments to the contrary.
In July 1983, IEEE 802 forwarded its token bus recommendation to the Technical Committee of Computer Communications (TCCC) for approval. Miller of Concord Data Systems remembers his role in the process:
We succeeded in getting token bus passed, so great, holy shit, a little company pushed that through. We pretty much wrote the standard. The other company that participated was Amdax, another token bus start-up.
Token ring, on the other hand, remained in limbo due to IBM’s vacillating leadership. In September 1983, they staged a technology demonstration of token ring in Zurich. Even so, no one expected a token ring standard recommendation to be out of committee until the spring of 1984.70
In September 1983, ISO SC6 met. As expected, the membership passed the recommendation that the IEEE 802 LAN standards be adopted. Little doubt remained that the recommendations would be approved by ISO at its next meeting in January 1984. The need for SC6 to create a connection-less network layer protocol was all that stood in the way for ISO/OSI to have a comparable LAN protocol stack to TCP/IP over Ethernet.71
One big question hovered over the emerging standards however: What was IBM going to do? (A decision referred to by some as: “the most speculated upon product of the decade.”72) The world’s dominant computer company had played its cards so close to its vest that it now seemed confused and oddly incompetent. Even without IBM’s stamp of approval, LANs had successfully run the gauntlet of international standards making. If IBM did not act soon, they might be left with the role of runner-up. Perhaps that motivated a cloaked presentation by Texas Instruments in February at the International Solid State Circuits Conference of a token ring chip set it had under development.73 If intended to quell rumors, it did little but fan them: chip problems held up IBM’s plans.74
On May 8, 1984, IBM made one of the most vilified product announcements of all time when it recommended the IBM Cabling System and pronounced its Statement of Direction.75 According to the Statement of Direction, IBM intended to implement a “star-wired token ring local area network using the IBM Cabling System within the next two to three years.” Implied was compatibility with the IEEE 802 standard. Far and away the most dominant computer company, long thought unassailable technologically, could only recommend how to wire ones’ facilities. And to trust their LAN solution would be available in two to three years. If ever an action had been meant to forestall market development, IBM delivered it that ignominious day.
While IBM shadow danced, the NBS-led efforts to stage an OSI-compliant LAN demonstrations at NCC progressed. In March 1984, GM and McDonnell Douglas held the first MAP users group meeting with 36 companies attending. Boeing, not to be left behind, announced that with the help of NBS it would lead the creation of an OSI protocol stack for technical and office environments, later to be named Technical and Office Protocols (TOP).76 That same month, March, NBS released information on the OSI Class 4 Transport (TP-4) compatibility testing programs under development. It identified 421 scenarios to test the 5,700 lines of C code needed to implement TP-4.77 (Another metric of the complexity of TP-4: an Intel spokesperson claimed it took 50 man-months to develop.78) The considerable commitment needed to prepare for NCC had narrowed the list of potential participants to fifteen companies.79
While general reviews of the NCC show held in Las Vegas, Nevada, on July 9-12, 1984, were mixed, all praised the NBS-led demonstrations. One review called NCC “by and large a ho-hum affair” with the exception of “two eye-opening demonstrations of open systems interconnection for local area networks.”80 The GM coordinated MAP over token bus (802.4) demonstration interconnected six devices from Allen-Bradley, Concord Data Systems, DEC, Gould, HP, IBM and Motorola. The more limited Boeing demonstration had files being swapped using TP-4 over 802.3 (CSMA/CD) between eight vendors: Advanced Computer Communications, Charles River Data Systems, DEC, Honeywell Information Systems, HP, ICL, Intel and NCR. The often-critical Data Communications editorial column, Viewpoint, ran the headline in August 1984: “Vendor-user partnership gives new meaning to standards.”81 The article ends:
We anxiously await such products from all the participating vendors and congratulate them for what we hope becomes a role model for continued vendor-user cooperation.
The workings of IEEE Committee 802, ISO, NBS, GM and now Boeing may have grabbed the headlines and chartering the future course of LANs, but the immediate driving force of LANs rested firmly with the IBM Personal Computer (IBM PC). In 1983, the growth rate of PC connections exceeded the overall market growth rate by 50%.82 The market for PC LANs was predicted to grow from $49 million in 1983 to $232 million in 1988.83 So when IBM introduced its second generation PC, the IBM PC AT on August 1, 1984, it is little wonder that they felt the need to also introduce a LAN.
On August 14, 1984, IBM, after years of waiting, rumors, predictions, obfuscation and stonewalling, finally introduced its first LAN: the IBM PC Network.84 Was it the feared token ring that leveraged the customers’ investment in the IBM Cabling System? No and no again. It was a 2 megabit per second CSMA/CD broadband network OEM’ed from the start-up Sytek.85 Promising to carry video, it smacked of a poor man’s Wangnet. Even so, experts welcomed IBM’s market entry with pronouncements such as: “IBM’s announcement will change attitudes in the end-user community, which has hesitated to adopt local area networks.”86 In August, AT&T introduced its first LAN, a slow version of CSMA/CD known as Starlan. It would be anything but a star.
IBM was not alone in publicly aligning itself with one LAN technology and then switching its choice.87 In the fall of 1984, the DOD, which had funded LAN development from the beginning, announced that it would adopt the ISO standards, implying that ISO protocols would supplant the existing DOD standard TCP/IP protocols.88 This decision was further supported in a National Academy of Sciences (NAS) study commissioned by the DOD. (Similar to the NAS study commissioned by the FCC to sort out telephony deregulation issues in 1969.)89
In September 1984, the IEEE approved the token bus standard (802.4). It was automatically adopted by ISO as an OSI standard.
On October 31, 1984, the momentum for OSI standards received another boost when NBS issued its first Federal Information Processing Standard 107 titled: Local Area Networks: Baseband Carrier Sense Multiple Access with Collision Detection Access Method and Physical Layer Specifications and Link Layer Protocol. It would become part of the government stack of protocols being developed by NBS known as Government OSI Information Protocols, or GOSIP. ISO’s OSI appeared the clear winner among computer communication protocols.
In October, IEEE 802 forwarded the token ring (802.5) standard recommendation to IEEE TCCC. On December 31, 1984, all four IEEE 802 LAN standards were approved by ANSI. ISO approved all four standards as well.
Data Communications, December 1983, p. 15
Around this time SC 6 took over responsibility for the transport layer as well and SC 16 was disbanded and a new SC 21 became responsible for the top three layers of the Reference Model.
Datamation, January 1984, p. 13
Electronics Week Sept. 17, 1984 p.17
Data Communications, April 1984, p. 16
Jean Barik, “MAP:A User revolt for standards,” Data Communications, Dec. 1985, pp.147-156
TOP, as was MAP, would be organized under the Society of Manufacturing Engineers (SME), for reasons similar to why other multi-company standard-making efforts were organized under the IEEE or NBS. The interested reader is referred to: Steven A. Farowich, “Communicating in the technical office,” IEEE Spectrum, April 1986, pp. 63-67
Data Communications, March 1984, pp. 277-285
Computerworld, July 23, 1984, p.52
Electronic News, July 9, 1984, p.
Electronics Week, July 23, 1984, p. 45
Data Communications, August 1984, p. 13
“Local Area Network Equipment,” Dataquest 1987, p. 77 and 80
Electronics Week, Dec. 10, 1984, p. 20 Total Lan sales exceeded $1 billion.
Technically, the PC Network was IBM’s second LAN. The first was the PC Cluster introduced in 1983 which had bombed. (DC, October 1984, p.77)
Ibid, pp. 20-21
“IBM’s first net well received; analysts call for more,” Computerworld, August 20, 1984, p. 5. The quote is from Evertt Meserve of Arthur D. Little.
Jean Bartik, “IBM’s token ring: Have the pieces finally come together?” DC, August 1984, pp. 125-139
Data Communications, Nov. 1984, p. 15
See Chapter 4