In the 60s and 70s telecommunications was far more primitive than it became in the 90s with deregulation privatisation and major advances in technology. Bit error rates could be lower than 1 in 1E5 in Europe. Leased line speeds were 9600bps or less. Costs in Europe were rather high. Public data networks did not exist.
I.P.Sharp Associates introduced a packet-switching system in 1976. The original objective was to provide remote access to a time-sharing host in Toronto. Nodes were mostly CAI LSI-2/20 minicomputers with IBM 3705s for host connection. About twenty nodes were installed by the end of 1976. By summer 1977 various hardware and software errors had been fixed and network reliability became quite good. In May 1981 when support for arbitrary topology was introduced there were about 110 production nodes including seven 3705s. Net90 technology based on the IBM PC/AT was introduced in 1986. In winter 1989 the total number of nodes was just under 256.
In the early 90s the size of IPSANET exceeded 300 nodes. In 1993 a decision was made by Reuters to replace it with newer technology from Hughes. The network was gradually dismantled. The Toronto datacentre was permanently shut down on 30 June 2005.
This site includes both archival material written in the 70s and 80s and new essays written in 2005. I have tried to make documents available in HTML form. Due to character set etc issues some documents are only available as PDFs. Some HTML documents assume a browser which supports Unicode.
Alpha&3705 Photo Gallery (about 250KB)
Photos of IPSA Datacentre and Remote installations
Essays written in 2005
Some new material has been created especially for this Website. These are essays written in 2005 mostly from memory and seem to have more errors in dates than in algorithms. Installation History is probably of the greatest interest to the casual reader.
IPSA and the Telephone Monopolies
This essay by Ian Sharp describes relations with sundry telephone monopolies in North America and abroad. In the 70s they were quite reluctant to accept technological change and attempted to retard it. Their restrictions made international e-mail illegal. This contrasts with the present situation where the only constraints on Internet applications are copyright and libel laws.
Alpha Hardware History.
The original CAI hardware configuration and its eventual evolution are discussed. Sundry hardware problems are belaboured in excessive detail. Two peripheral boards constructed especially for IPSA by Macrodata BV are discussed.
3705 Software functions (other than normal network protocols) are described including AWAITING FUNCTION FOUR.
I3L Language and the compiler and link editor which implemented it are discussed.
Framing and Data Link Control protocols of IPSANET are details. Parts of the data link control exposition assume a limited knowledge of programming.
Installation History. This is a history of the network emphasizing installation of nodes and links rather than protocol and implementation details.
Interend Protocols which were used by some or many customers are described. T-task, Bisync, dialout, SDLC and X25 are mentioned. Some protocols are described in medium detail. Bandwidth efficiency is analysed.
Network Management Protocols describes the Down Line Load protocol used to load an Alpha from the mainframe and the logging task which ran on the mainframe.
Route Control revisits some of the material from the 1981 Newsletter article on the new routing protocol. More analysis, some anecdotes and other material are included.
I.P.Sharp Associates published various newsletters to service different customer groups. The Newsletter had the broadest circulation and went to employees, customers, friends and competitors. INSITE was a narrowly focused newsletter address to clients who were running the Sharp APL software on their own mainframes. About a third of these were connected to IPSANET.
Newsletter March/April 1978 Lead article by Geoff Lewis on the network. Article is brief but fairly accurate.
Newsletter September/October 1978 Lead article by David Chivers announcing Telex access to IPSANET. Article discusses this new facility from a user view including benefits and limitations.
Newsletter January/February 1983 Four page article by Paul Berry was published after 3705 error message format was changed. The article explains the connection and signon process and explains the various places which might fail and generate an error message. Messages from Alpha, 3705 and APL are documented
IPSANET Manual March 1984 24 page manual which provides much information on how to connect an asynchronous terminal to IPSANET. Other services are given multi-paragraph descriptions. NOTE: Accurate viewing of Section 3.3 of this manual requires support for CSS2 @font-face (Firefox or Safari browser) to display APL characters. Correct alignment of the column data in Section 3.3 requires Internet Explorer or Opera.
INSITE September 1986 2nd page article IPSANET: Yesterday, Today and Tomorrow by David Chivers. This is a good summary of past achievements and plans for Net90.
INSITE September 1986 Using 3270 Applications through IPSANET by David Chivers. The SDLC pipeline and its use by IPSA are described.
INSITE September 1986 The IPSANET X.25 Gateway by Malcolm Prescott. Various uses of the X.25 Gateway are described. A brief note on the X.25 PAD has been appended.
INSITE January 1987 The Future of IPSANET by David Chivers. This outlines sundry hopes for Net90.
This winter 1989 e-mail was sent to all internals. The Alpha is declared obsolete. It discusses some user visible changes to the network resulting from the introduction of the Beta node (from Net90 project). The ability of the Beta to connect to X.25 networks as well as APL systems required some changes in users' initial connection procedures.
SIN-26 Bulletin to inhouse customers describing the IBM 3705 hardware configuration required to connect to IPSANET.
April 1977 to March 1978: e-mail messages from network correspondence file which mention future installation of the Vienna node. Collection ends shortly before the node become operational. This is raw history.
Three longer essays by Roger D Moore from 1978, 1981 and 1982
An APL Users Meeting September 1978: A DATA COMMUNICATIONS SYSTEM FOR THE APL USER. Six-page article discussed various aspects of the network in some detail. Design objectives are mentioned and arguments are made for the technology used within the network.
Newsletter TS July/August 1981: MAJOR NETWORK CHANGE. The switch to an all virtual-call datagram-free routing scheme is described with some attempt at mathematical rigour.
1982 APL Users Meeting Proceedings Volume I Applications. NETWORK MANAGEMENT TOOLS describes various APL programs used in the operation and support of the network. This eight page paper is an overview and not particularly formal.
This paper by Richard L Potyok was published by the Association for Computing Machinery in Quote Quad Vol 18, no 3.
The Network Shared Variable Processor extended APL shared variables to remote host computers. This eight page paper describes the 370 NSVP program. User facilities and some internals are discussed. 3705 and X.25 Gateway aspects of NSVP are briefly described in the Interend Protocols essay.
These papers were written by I.P.Sharp customers. They describe actual experience with IPSANET in writing applications for use by non-programmers.
1980 APL Users Meeting USE OF TELEX FOR FINANCIAL PLANNING & REPORTING By Roger W. Shaw, The Wellcome Foundation Ltd. describes two applications which used the Telex interface to Sharp APL.
1980 APL Users Meeting Private Networking in a Multinational by Irene P. Hartford, Massey Ferguson Limited describes MF's private network. Some services were provided via IPSANET technology. In 1973, a half-speed trans-Atlantic Telex link met their telecommunication needs. By 1980 demand had grown substantially.
1982 APL Users Meeting Developing APL Systems at the Urbain Life Re by Michal I. Bauer does not say much about IPSANET. It illustrates the problems of trying to use a terminal protocol for file transfer. (3270 VTAM users had similar problems.)
Some Marketing Brochures pertaining to IPSANET are presented here. All four were written by Mary Lee Coombs.
February 1986 IPSANET The Choice is Clear
It contains one paragraph descriptions of eight different customer applications. A list of branch offices is also included. Two page map has been omitted.
February 1987 IPSANET: Are You International?
Five applications (some the same as Feb 87 list) are mentioned.
July 1987 IPSANET Directions
This appears to be a condensation of INSITE September 1986 article entitled IPSANET: Yesterday, Today and Tomorrow
July 1987 IPSANET & Data Security
Some claims made for IPSANET refer to rather specialized situations.
Glossary defines some telecommunications terms plus jargon peculiar to either IPSANET or IBM.
Network Topology diagrams were published in the IPSA newsletter on a regular basis for several years. The topology eventually became more complex than the primitive program for generating a diagram from the pairwise description of the network could handle. Unfortunately it was never repaired or replaced and this valuable source of network history ends in August 1981.
Documents have been provided by: Eugene McDonnell, Jane Minett, Tony Bailey, Leslie Goldsmith, David Chivers, and Michael Harbinson
Criticism and corrections been provided by Fred J Perkins, Michael Harbinson, Joey Tuttle, Dave Markwick, Eric Iverson, David Chivers, Robert Bernecky, Ian Sharp, Larry Breed and Peter Wooster.
Technical support was provided by Rosanne Baumhard and Tony Bailey
Research assistance was provided by John Thompson