William Stallings

The history of the development of the OSI model is, for some reason, a little-known story. Much of the work on the design of OSI was actually done by a group at Honeywell Information Systems, headed by Mike Canepa, with Charlie Bachman as the principal technical member. This group was chartered, within Honeywell, with advanced product planning and with the design and development of prototype systems.

In the early and middle '70s, the interest of Canepa's group was primarily on database design and then on distributed database design. By the mid-70s, it become clear that to support database machines, distributed access, and the like, a structured distributed communications architecture would be required. The group studied some of the existing solutions, including IBM's system network architecture (SNA), the work on protocols being done for ARPANET, and some of the concepts of presentation services being developed for standardized database systems. The result of this effort was the development by 1977 of a seven-layer architecture known internally as the distributed systems architecture (DSA).

Meanwhile, in 1977 the British Standards Institute proposed to the International Organization for Standardization (ISO) that a standard architecture was needed to define the communications infrastructure for distributed processing. As a result of this proposal, ISO formed a subcommittee on Open Systems Interconnection (Technical Committee 97, Subcommittee 16). The American National Standards Institute (ANSI) was charged to develop proposals in advance of the first formal meeting of the subcommittee.

Bachman and Canepa participated in these early ANSI meetings and presented their seven-layer model. This model was chosen as the only proposal to be submitted to the ISO subcommittee. When the ISO group met in Washington, DC in March of 1978, the Honeywell team presented their solution. A consensus was reached at that meeting that this layered architecture would satisfy most requirements of Open Systems Interconnection, and had the capacity of being expanded later to meet new requirements. A provisional version of the model was published in March of 1978. The next version, with some minor refinements, was published in June of 1979 and eventually standardized. The resulting OSI model is essentially the same as the DSA model developed in 1977.

Copyright 1998