General ATM Questions

Cell Relay Retreat>ATM FAQ> General

The ATM FAQ is maintained as part of the Cell Relay Retreat project
sponsored by One Call Internet, Indiana University, and Werner Digital Technology.
  1. Who is the inventor of ATM?
  2. How does ATM map to OSI layers?

Who is the inventor of ATM?

This topic is about who was the "father", or "mother", or "parent" of ATM rather than inventor so to speak. Three names generally come up as an answer depending on the perspective.
  1. Jon Turner
  2. J.-P. Coudreuse
  3. Sandy Frasure
Jon Turner (at Washington University (wustl.edu)) is to some the "parent" of ATM as it has become deployed in the enterprise and ISP markets. This is in contrast to the "heavy iron" central office switches which started development first. Note this perspective does not view him as the inventor of ATM, rather he has been viewed by some as a motivating force behind ATM in the non-CO market. He was writing papers on ATM way back into the early 80's. His student team developed an initial chip set which was actively marketed to corporations (e.g. Synoptics) for the development of small ATM switches. The subsequent marketing splash of this and the products of other companies gave inertia to the enterprise and LAN ATM switch market. That then showed up on the radar screens of the weekly trade rags and then the next thing you know a "parent" is proclaimed (as such in the press).

Others focus on the question of who constructed the first practical ATM demonstration. Some suggest that this person is J.-P. COUDREUSE, one of the technical managers of the CNET "Prelude" project in the early '80s. "Prelude" was apparently the first project that practically demonstrated the feasibility of service integration with cell switching, encompassing *both* networking and application aspects.

Others focus on the earliest researcher for cell-based ATM like technology. Sandy Fraser has been mentioned to be the earliest. He had worked with J.-P. COUDREUSE in the early '80s but began ATM research in the early 70's. Fraser has presented an excellent summary (in the January 1993 issue of IEEE Network) of the early Bell Labs work on Asynchronous Time Division Multiplexing, easily recognized as cell-based ATM. One product based on this work is the Datakit Virtual Circuit Switch, commercially introduced in December 1983 and still in use by various organizations formerly known as the Bell System. The early research started in 1969, about the time of Aloha and BBN's ARPA IMPs.

The IEEE Network article shows a fairly clear evolution of the idea of small fixed-sized cells with circuit identifiers (VCIs) in headers used to switch cells to appropriate outputs. The need for cell header translation is clearly shown. One application of the design in 1977 was called Incon, with 16-byte-cells that supported both voice and data, and a congestion scheme giving priority to isochronous traffic.

How does ATM map to OSI layers?

Most people agree that the ATM standards cover 3 distinct layers -- Physical Layer, ATM Layer, and ATM Adaptation Layer (AAL).

The Physical Layer (corresponding to OSI Physical) is usually taken to be SONET/SDH (which itself has 4 layers...) but can be other things as well. The PHY deals with medium-related issues.

The ATM Layer is responsible for creating cells and formatting the cell header (5 octets). Some argue that it also corresponds to OSI Physical (it deals with bit transport) and others say its OSI Data Link (formatting, addressing, flow control, etc.).

The AAL is responsible for adapting ATM's cell switching capabilities to the needs of specific higher layer protocols. The AAL is responsible for formatting the cell payload (48 bytes). Some argue that this layer corresponds to OSI data link (data error control, above Physical), others OSI transport (it's end-to-end).

I think that this all proves that the OSI model is an excellent model as a basis for discusion and comparison but does not have a clear mapping to ATM.

    









© 2000-2004 Cell Relay Retreat