AM 254 Remembered

[The following notes were written in July of 2017.]

For Harvard's spring term of 1975, John McQuillan and Dave Walden led a graduate seminar on computer networks, numbered Applied Math 254. Perhaps a couple of dozen (maybe less) graduate students (and some undergraduates) from Harvard (and cross-registered from MIT) participated in the seminar.

McQuillan and Walden were part of ARPANET team at Bolt Beranek and Newman (BBN), in Cambridge, MA, two miles west of Harvard. McQuillan had recently finished his PhD at Harvard (while working at BBN) with adaptive routing in computer networks as his thesis topic. Professor Tom Cheatham, who was interested in practice as well as theory, invited McQuillan to give the graduate seminar, and McQuillan invited Walden to co-lead the seminar.

This web page displays the handouts for the seminar which, in July 2017, were borrowed and scanned from Guy Steele, one of the students in the seminar.

These handouts are worth displaying for two reasons:
(1) the seminar was an early college course in packet-network design, perhaps the first led by implementors of an operational packet network.
(2) the handouts give a snapshot (at least from McQuillan and Walden's point of view) of the state of computer networking at the end of 1974. This was one of the early turning points in networking, as the ARPANET had reached technical and operational maturity, but the Internet was just getting started, along with commercial packet switching and international efforts.

The rest of this cover page provides a table of contents and commentary on the handouts. The handouts are listed below in their approximate order from Steele's three bound packets of the handouts, which we assume was the order of handout in 1975 as Steele is a rigorous archivist. Some of the division of the handouts into class sessions from class 8 and beyond is supposition based on the syllabus as the handouts don't include class session indications beyond class 8. Perhaps, some papers were handed out at a later or earlier class session than they were actually relevant to. Various of the papers were relevant to more than one session.

If anyone else who participated in this seminar finds this web page and has any memories of the seminar, please pass your thoughts on to dave@walden-family.com.

Course syllabus

Notes for Classes 1 and 2: Introduction to Computer Networks

As this web page is being written in 2017, the guess is that despite what the syllabus said for classes 1 and 2, the above 115 page Introduction to Computer Networks and the long list of other handouts below probably took two classes to cover (and in so doing somehow covered the syllabus topics for both class 1 and class 2).

Networks: An Introduction by David Farber.

Topological considerations in the design of the ARPA computer network by Howard Frank, Ivan Frisch, and Wushow Chou.
Only the first page was scanned from the 1975 documents to save scanning; the full paper was pulled from Internet in July 2017.

Data by packet by Lawrence Roberts.
Only the first page was scanned from the 1975 documents to save scanning; the full paper was pulled from Internet in July 2017.

Origin, Development and Current Status of the ARPA Network by Peggy Karp.
Only the first page was scanned from the 1975 documents to save scanning.
Citation: Peggy M. Karp, Origin, Development and Current Status of the ARPA Network, in SEVENTH ANNUAL IEEE COMPUTER SOCIETY INTERNATIONAL CONFERENCE, DIGEST PAPERS 49-52 (1973).

Analytic and simulation methods in computer network design by Leonard Kleinrock.
Only the first page was scanned from the 1975 documents to save scanning; the full paper was pulled from Internet in July 2017.

Network Performance excerpt from McQuillan's Harvard thesis on Adaptive Routing Algorithms for Distributed Computer Networks.

The Network Control Center for the ARPA Network by McKenzie, Cosell, McQuillan, and Thrope.
Only first page scanned from 1975 documents to save scanning; the rest of the paper was pulled from Internet in July 2017.

On Measured behavior of the ARPA Network by Leonard Kleinrock and William Naylor.
Only the first page was scanned from the 1975 documents to save scanning; the full paper was pulled from Internet in July 2017.

The ARPA Network Control Center by Alexander McKenzie.

Reliability Issues in the ARPA Network by William Crowther, John McQuillan, and David Walden

The Satellite IMP for the ARPA Network by S. Butterfield, R. Rettberg, and D. Walden.

Presentation and Major Design Aspects of the Cyclades Computer Network by Louis Pouzin.

Software checksumming in the IMP and network reliability by John McQuillan.

Notes for Class 3: Definitions and Fundamental Properties

The title of the notes linked to in the above line is different from the title in the syllabus (Network Performance).

Slides/Notes for Class 4: Computer Network Topology

This class was presented by Mario Gerla, who at the time was at Network Analysis Corporation (NAC) on Long Island in New York State. Gerla had received his PhD at UCLA, working on the early ARPA Network system and protocols under the guidance of Prof. Leonard Kleinrock. After a few years at NAC, he returned to UCLA where he has been a professor ever since. NAC did the topological layout analysis for the ARPANET.

This class was one of two instances in the course where McQuillan and Walden invited a guest speaker to present one of the class sessions, in order for seminar participants to hear from other experts.

Notes for Class 5: Node-to-node Transmission Algorithms

Notes for Class 6: Source-to-Destination Transmission Algorithms

The Influence of Control Procedures on the Performance of Packet-switched Networks by Opderbeck and Kleinrock.

Basic Elements of a Network Data Link Control Procedure by L. Pouzin.

A Protocol of Packet Network Interconnection by Cerf and Kahn
Only the first page was scanned from the 1975 documents to save scanning; the full paper was pulled from Internet in July 2017..

One or more of the following three papers may be follow-up handouts from class 4 or belong with class 7 instead of 6.

Cost Throughput Trends in Computer Networks using Satellite Communication by Gerla, Chou, and Frank.
Only the first page was scanned from the 1975 documents to save scanning.
Citation: "Cost Throughput Trends in Computer Networks Using Satellite Communications", Mario Gerla, Wushow Chou, H. Frank, in International Conference on Communications (ICCC), pp. 21C-1 to 21C-5, Minneapolis, USA, June. 1974.

Optimal Routing in a Packet-switched Computer Network by Cantor and Gerla.
Only the first page was scanned from the 1975 documents to save scanning.
Citation: M. Gerla, D.G. Cantor, "Optimal Routing in a Packet-Switched Computer Network", IEEE Transactions on Computers, vol. 23, no. , pp. 1062-1069, October 1974, doi:10.1109/T-C.1974.223806.

Chapter 6, Topological Design — apparently from a draft of a book or from a report, perhaps authored by people from NAC.

Notes for Class 7: The Routing Algorithm

Notes for Class 8: The Communications Processor

The handwriting for the above notes on the communications processor is McQuillan's.

Three papers on BBN's multiprocessor packet switch.
- The BBN Multiprocessor by Ornstein et al.
- The Evolution of a High Performance Modular Packet Switch by Ornstein and Walden.
- Pluribus — A Reliable Multiprocessor by Ornstein et al.

We are not sure of the exact order of the next several planned class sessions. They may have been reordered to accommodate the schedule of our second guest speaker.

Handouts for Class 9: Network Applications and Basic Protocol Issues

The paper in class 10 below would also have been used in this class.

An RFC by Bressler, Murphy and Walden about a proposed experiment with a different type of host-to-host protocol, for comparison with the kind of protocol described in document above.
A paper from the Communications of the ACM by Walden describing the different type of host-to-host protocol.
Neither the different protocol nor experiment went anywhere.

Handouts for Class 10: Host Protocols

Pre-publication version of a Walden presentation on Host-Host protocols.
Published version of the same presentation (not part of the 1975 seminar handouts).
The pre-publication version is more in the author's voice.

See the syllabus for other documents that were surely drawn from for this class.

Handouts for Class 11: Political, Economic, and Social Issues

Stuart Mathison was our second guest speaker. In 1970 he and Phillip Walker co-authored the book Computers and Telecommunications: Issues in Public Policy, and in 1975 he was the vice president for planning for Telenet Corporation.

A brochure and two newsletters about Telenet's activities.

Matters before the FCC, docket 19746.
Starting in 1973, value-added common carriers began to be approved. These are carriers that don't construct their own communications links but lease lines from other carriers and create their own telecommunications services using computers that communicate via the leased circuits. Telenet was a value-added carrier.

Handouts for Class 12: Internetworking and New Technologies

The Cerf/Kahn paper listed with class 6 was reused with this class.

Packet Radio Station Hardware, Operating System, and Applications Programming Environment by Burchfiel, Tomlinson, and Beeler.
The published version of the above paper

The Atlantic Packet Satellite and Gateway Experiments by Binder, Rettberg, and Walden.

Classes 13 and 14

One of the last sessions of the seminar involved a trip to BBN for the class. A class session or two earlier, McQuillan and Walden asked the students what other topic or guest speaker they would like to hear about or from. Class members said that wanted to take a field trip to BBN to see the ARPANET developers and operators in action and they had already made arrangements to have a Harvard Radcliff shuttle bus take the seminar participants there. (It also seemed that some of them were looking for summer or full time jobs. In fact, BBN hired two or three of the participants from the seminar, although recruiting had not been an idea McQuillan and Walden had in mind when they planned the seminar. Still, it was not unusual for BBN to get to know students before they graduated from Harvard and MIT, perhaps through faculty members who consulted to BBN. In fact, McQuillan's own first connection to BBN was through a Harvard course taught some years earlier by BBN engineer Severo Ornstein.)

Perhaps the other of the last two class sessions involved answering additional questions from the participants and passing back term project papers. McQuillan and Walden reviewed the term papers from class members while traveling to the AFIPS Joint Computer Conference in Anaheim in May 1975, probably between the last two class sessions. At the conference they presented a paper about topics that had been presented in the course.

A couple of concluding notes

There were several relevant papers on the ARPANET by BBN authors which apparently we did not hand out; presumably our class presentations were sufficient to cover that material. Examples are the 1970 AFIPS Spring Joint Computer Conference paper by Heart et al. and the 1972 AFIPS Fall Joint Computer Conference paper by McQuillan et al. Also, we covered the ARPANET Network Control Protocol and other early protocols even though we did not hand out some key early papers about them, e.g., the Carr et al. paper from the AFIPS 1970 Spring Joint Computer Conference or the Crocker et al. paper from the AFIPS 1972 Spring Joint Computer Conference.

The files linked to by this page are in order (in their directory) by file name which matches the order in Guy Steele's hard copy of the course handouts.