1970-1980

De Pssstopedia.

Si le web a vu le jour sous Internet, l'impact de la transmission des données sur le grand public a commencé bien avant. Dès 1976, Montréal était l'hôte d'un nœud du réseau Datapac qui a servi durant 30 ans à la vérification des cartes de crédit et de débit au profit de millions d'utilisateurs sans qu'ils s'en rendent compte. C'est d'ailleurs Datapac qui me permettait d'avoir accès à INFOPUQ à partir de mon Commodore 64 en 1983. Voici l'introduction d'un article paru récemment qui relate les tout premiers débuts des réseaux de données publics.

CONTEXT In the 1970s packet networking research was vibrant around the world. Larry Roberts [1], who many call one of the fathers of the Internet, discussed many of these in his article. I would highlight ARPAnet in the United States, Cigale and the Reseau de Commutations des Paquets (RCP) testbed in France, and work at the U.K. National Physical Laboratory. Most of this work was built on commercial minicomputers of the day (e.g., DEC PDP11s). A small number of commercial store-and-forward message-based services had been developed, for example, one run for the airlines by the Societé Internationale de Télécommunications Aéronautiques (SITA), and one in Spain operated by Telefonica for interbank networking. There was also Tymnet in the United States, which multiplexed character streams from multiple users in addressed transmission blocks. Enterprise networking was in its infancy, based on star networks run over leased lines. These were typically mainframe-based with remote job entry (RJE) and IBM 3270-style access, using vendor-specific protocols defined by IBM and the other computer manufacturers (e.g., Digital). General-purpose user access was provided by “dumb” async terminals (e.g., VT100) via lowspeed modems over the public switched telephone network (PSTN). At the time, T1/E1 leased lines were very expensive. There were no PCs (the Apple II debuted in 1977, the IBM PC in 1982, and the Apple MacIntosh in 1984!). Commercial LAN deployment would only start in the 1980s. IP was a research tool. In 1972 Bell Canada, on behalf of the TransCanada Telephone System (TCTS), set up a dedicated engineering, operations, and marketing organization (called the Computer Communications Group, CCG, and referred here to as Bell/CCG) to develop advanced data services. In November 1974 Bell/CCG announced plans to introduce packet switching services by July 1976. In the United States, Telenet had launched proprietary packet network services in 1975 and saw the value of an international standard. The French post, telegram, and telephone agency (PTT) were going down the path of rolling out what became Transpac (announced in 1973), and Euronet on behalf of the European Economic Commission (EEC). The British Post Office was running their Experimental Packet Switching Service, and NTT in Japan was likewise researching packet networking opportunities. There was one major problem with attempts to commercialize packet switching. Without a global standard, IBM (and the “seven dwarfs,” as the other seven smaller computer manufacturers were called) would not have a strong business case to connect to these networks, which in turn heavily impacted the business case for these carriers to launch services.

THE DRIVE FOR GLOBAL STANDARDS AND SERVICE ALIGNMENT Bell Canada owned Northern Electric (later called Northern Telecom, Nortel Networks, and Nortel) at the time, and together they owned Bell-Northern Research (BNR). BNR had a packet switching research group and was a strong partner in Bell/CCG’s packet project. We saw the advantages of packet switching over circuit switching summarized in Table 1. Bell/CCG issued a draft packet interface specification proposal called the Standard Network Access Protocol (SNAP). SNAP included three layers on top of the physical layer: 1. The data link control layer 2. The datagram layer 3. The virtual call procedure layer, which included a window-based flow control/error recovery/ sequencing/duplicate detection mechanism. The datagram/virtual call procedure layers had many of the elements of TCP/IP, outlined in a 1974 paper by Cerf and Kahn [2]. The draft proposal was sent to carriers and hundreds of researchers in the preeminent laboratories around the world, and of course to IBM. Our objectives were (1) to kick start the discussion of the business need for a global standard (ultimately called the X.25 family of standards) and alignment of public service development timelines; and (2) to propose the key attributes of the required standard.

"Commercialization of Packet Switching (1975–1985): A Canadian Perspective", Tony Rybczynski, IEEE Communications Magazine, Vol. 47, no. 12, December 2009.