Philip Kelly Computer communications the early days 1966 – 1986 Part 2 As outlined in the first part of Philip Kelly’s article [1], the British Post Office (BPO), had decided not to implement a dedicated circuit-switched data network but to await the digitalization of the UK’s network to provide circuit switched services. In the second part, further information is given on the early development of packet-switched data networks Experimental packetswitched networks By 1970, George Allery’s staff within the BPO’s development department had begun to prepare specifications for the design and implementation of a data network based on the packet mode of working. Much of the details relating to actual protocols were developed by Mike Smith and his team [2, 3]. Around the same time, starts had been made on other experimental 54 | Volume 5 Part 1 • 2011 that enabled the International Telegraph and Telephone Consultative Committee (CCITT) to finalise the X Series of Recommendations. He also examines the events leading up to the establishment in the UK of a public data network based on the packet mode of working. An indication is also given of developments in the telex and teletex areas. networks. In France development had started on their Réseau à Commutation de Paquets (RCP) [4]; in the USA the Advanced Research Projects Agency Network (ARPAnet) was being designed [5] and in the UK the NPL, together with other research institutions, were involved in developing the European Informatics Network (EIN) [6]. There were also other experimental networks being developed such as Xnett in Norway, one in Canada by Bell Northern Research and one in Japan by the Nippon Telephone and Telegraph. In the early days, there was still considerable doubt about how a packet-based data transmission service should be implemented and as to what charges for its use should apply. Therefore, the board of the BPO decided that such a service should only be introduced on the basis of an experimental national network for THE JOURNAL OF THE INSTITUTE OF TELECOMMUNICATIONS PROFESSIONALS joint evaluation by both the BPO and by users and terminal manufacturers. Approval for the implementation of a trial network was not given until 1973 and I was appointed to oversee all aspects of its implementation and operation. The Experimental Packet Switched Service (EPSS) was eventually fully opened for service in 1977, using Ferranti Argus 700E processors [7] – see Figures 1 and 2. Photo courtesy of Martin Baker There was close co-operation at the time with those experts who were working elsewhere on experimental networks and there was a considerable interchange of information on the specification of protocols that needed to be developed. The link level and other protocols, which were developed by Mike Smith and his team, were implemented in EPSS but they were extremely complicated and were proving to be difficult and expensive to realise and it is believed this delayed the opening of the service. The concept of “virtual calls” and “permanent virtual circuits” and the idea of combining packet transmission and a connectionbased service were found to be of considerable appeal, especially to the French. The various presentations and discussions eventually led to the establishment of agreed standards for a Link Access protocol. This was based to some extent on work done within the International Standards Organisation for a balanced High Level Data Link Procedure (HDLC) and in addition took account of the preference in the USA for a master/slave protocol. Figure 1: EPSS equipment 110bit/s or 300bit/s Character terminals Character and packet terminal arrangements as for Manchester 110bit/s or 300bit/s Glasgow 48kbit/s links 110bit/s or 300bit/s Manchester 48kbit/s links 2400bit/s or 4800bit/s Packet terminals 48kbit/s links 48kbit/s Character and packet terminal arrangements as for Manchester Packetswitching exchange London Connections via PSTN Data terminal equipment Direct connections Figure 2: EPSS schematic at its opening in 1977 Eventually the CCITT, via its special rapporteur group for the packet mode of working, recommended an agreed Link Access Protocol Balanced (LAPB) and this was incorporated into Recommendation X.25 and many other data link protocols for services being implemented during the 1980s. EPSS was basically a test bed. It was of particular value in enabling the data processing industry, users and the BPO to evaluate the technical problems of implementing packet switching protocols in computer terminal equipment. Other countries followed the development of EPSS with considerable interest. From EPSS much was learned about the concept of layered modelling of protocol structures, (network, transport, virtual terminal, file transfer etc.) together with practical experience in attempting open systems working between different computer systems. All the time EPSS was being developed, the CCITT continued to refine its Recommendations for packet-based services. Enhancements were made to X.25 and, within the Conference of European Postal and Telecommunications Administrations (CEPT) drafts for what eventually became CCITT Recommendations X.28 and X.29 for character-based terminals were developed under the chairmanship of John Wedlake who eventually took over as Chairman of the CCITT Study Group VII in 1986 and continued in this post until 1993. EPSS was never expected to be commercially viable as no charges for its use were to be levied; only dummy bills would be issued and these were based on the number of packets transmitted. EPSS was soon overtaken by the development of international standards for user terminal/packet switching network interfaces. In particular the issue by the CCITT of the X.25 Recommendation in 1976 had a profound effect on developments as did advancements in microprocessor technology. EPSS was not compatible with the Recommendations in the X series particularly X.25 and it was not considered economical for it to be modified. It was in due course withdrawn from service. Charging principles and tariffs Even though it was not intended to charge for the use of EPSS, it was considered that an early start should be made on determining the charging structure that should apply to networks based on the packet mode of working. Tariffs for telecommunications services needed to be set by the BPO’s service department that had, at the time, little or no understanding of computer communications and were only really interested in obtaining the appropriate rate of return on the actual investment. The tariff division was headed by Ena Knight who had graduated with an engineering degree Volume 5 Part 1 • 2011 | 55 but could not join the BPO’s engineering department which was, at the time she graduated, a male-only preserve. She thus joined the BPO’s traffic department dealing with service and policy matters. Following pressure brought by the marketing department and myself, the charging rules for packet-based services were agreed in principle and were to be quite different from those for conventional telephony services. There had already been some discussions on the subject of charges, initiated by Larry Roberts of Telenet in the USA, with Dave Horton of Bell Canada, Philippe Picard of France and myself. It was proposed by this group that, in spite of the various geographical sizes of countries, access charges should be dependent on the access data signalling rate but independent of the distance of the terminal from its packet switching exchange. A charge should however be levied for the number of packets transmitted. The CCITT issued in 1977 Recommendation D.10 relating to general tariff principles (Orange Book Vol II.1). These principles were accepted by Ena Knight for the UK network and thus call patterns were monitored on EPSS and nominal bills prepared. The proposal that charges should be independent of distance seemed to reflect the postal universal charge which perhaps was appropriate for “packets”. The Eurodata Foundation Whilst the International Computer Communications Conferences (ICCCs), to which considerable reference has been in given in Part 1 of this paper [1], enabled there to be discussions between technical people, there was little attention to the marketing or charging aspects for new data services. In 1971, Torsten Larson, the then Deputy Director of the Swedish Telecommunications Administration called a small meeting in Stockholm at which both Halvor Bothner-By, the CCITT Special Rapporteur for the packet mode of working and myself were present, to consider a proposal by American consultants to undertake a market study on data communications in Europe. Following this meeting CEPT Administrations jointly invited international tenders for such a study and this led to the publication 56 | Volume 5 Part 1 • 2011 of the “Eurodata 1974 Study” [8]. This study indicated a significant increase in the number of data terminals over the next decade and to the need for international interworking but the study did not take into account the existence of specialised data networks. In 1976 an Eurodata Foundation was formally established in London with members drawn from the CEPT Administrations. Its main role in the early days was to provide a forum for members to exchange information on the telecommunications market. In 1979, the study of tariff information became the focus of activity, especially the tariffs for public data networks. In 1981 it issued its 4th edition of a report on the plans of the Administrations for public data networks which by then included tariff information. In 1999, the Foundation became Teligen Ltd and continues to supply information relative to telecommunications services and the tariffs which apply. The National Committee on ComputER Networks 1976 As often happens in Government or semi government departments when things are uncertain, a committee is set up with loosely defined terms of reference to consider and report back in, say, a year or two, by which time the situation might have resolved The tariff division was headed by Ena Knight who had graduated with an engineering degree but could not join the BPO’s engineering department which was, at the time she graduated, a male only preserve. She thus joined the BPO’s traffic department dealing with service and policy matters. itself. In 1976 in the UK, the Secretary of State for Industry, Eric Varley, no doubt as a result of considerable pressure from scientific institutes, manufacturers and users, decided in true “Yes Minister” style to set up such a committee to be known as the National Committee on Computer Networks (NCCN) to consider the development of computer networks in the light of national economic and social factors, taking into account the joint interest of users, hardware and software suppliers, standards and the relationship between private and public networks. It was headed by Dr. Jack Howlett from the Atlas Laboratory. Of the fifteen members it is interesting to note it included Donald Davies, and my successor as the head of the Civil Service Department’s computer division Ray Atkinson, Frank Thomas the then director of BPO’s network planning department, and industry representatives such as Peter Hermon from British Airways. Bryan Stanley from the Post Office Engineering Union was also appointed. On occasions I deputised for Frank Thomas and provided him with detailed technical and commercial support. Surprisingly NCCN produced a report within two years [9] concluding and recommending amongst other things: • The need for a public switched data transmission service with international connections was seen as important and urgent. • A service using packet switching should be introduced as the first step. • The service should be provided by the BPO as part of the national telecommunications services and the tariffs should be distance-independent within the UK and pitched at an attractive level. • The BPO’s plans recently announced for a Packet Switched Service (PSS) were welcomed but it was noted there was no firm commitment and the planned rate of growth was considered to be on the low side. • The Department of Industry should consider contributing towards the cost and should also consider underwriting any start-up losses. • The BPO should improve its customer relations and give increasing advice regarding the problems of THE JOURNAL OF THE INSTITUTE OF TELECOMMUNICATIONS PROFESSIONALS connecting to the new network. Its monopoly in the provision of modems for connections to the PSTN should be removed. • Subscribers should be free to connect any approved equipment to the Data Communications Equipment (DCE)/ Data Terminating Equipment (DTE) interface and the procedure for approving attachments should be simplified. • The policy regarding interworking between public and private networks should be modified to encourage inter-communications. • Message switching as a private commercial venture should be licensed on a limited basis. The report was in general accepted by those involved but reservations were expressed in the report by Frank Thomas on behalf of the BPO and by Bryan Stanley on behalf of the unions. Towards the end of NCCN’s work, the BPO had actually announced plans for a 9-node national PSS in full accordance with the relevant CCITT Recommendations to supersede EPSS. Opening was expected to be in 1979. Frank Thomas felt that the planned growth for PSS was very flexible and adaptable to any demand. On the other hand Bryan Stanley on behalf of the Post Office Engineering Union could not agree to proposals regarding:• The relaxing of any rules regarding what could be attached to the network • The ability of private networks to be able to interconnect with public networks • That the BPO’s interest should end at the DCE/DTE interface. NCCN’s report more or less laid out a firm plan for the provision of a national packet-switched data service and it was interesting to note the acknowledgement it effectively gave to the small group of people in the BPO whose efforts to provide new services were often frustrated at every stage of implementation. As is mentioned later PSS did not in fact open in 1979 as announced but in 1981, mainly due to contractual and implementation problems. The Euronet/DIANE project The Commission of the European Communities (CEC) had set up a Directorate General XIII within which a Copenhagen Dublin Amsterdam London Frankfurt Brussels Luxemburg Paris Packet switching exchange Remote access facility Network management centre Rome 9600bit/s link 48kbit/s link Figure 3: Euronet at its opening in 1978 department under Dr Georges Anderla was to consider the need for making information readily available throughout the then existing Community. An international information network, based to some extent on the concepts used in the networks set up by the Airlines (Society International for Telecommunications Aeronautic) and by the Banks (Society for Worldwide Interbank Financial Telecommunications) was proposed. Telecommunications Administrations felt this was too big a concept and urgent meetings were held with the Commission at which it was agreed that the actual communications network would be provided by the Administrations and that the Commission would take responsibility for the information to be transmitted over the network. Thus the concept of Euronet, and the CECsponsored database access service which would use Euronet, was born. This latter service - Direct Information Access Network, Europe - was given the acronym DIANE, thus clearly showing the split between communications and the information accessible over Euronet [10,11]. As vice chairman of the CEPT’s data communications group, I took on the part-time task of overseeing the implementation of the network. In 1976 I was promoted to become the deputy director in charge of the planning of the trunk network other than the transmission elements and thus gave up my national responsibilities for data transmission. However it was agreed that I could continue my oversight of the Euronet project because of its significant implications for the future of international data transmission. A small project team under the leadership of Jean-Yves Gresser from the French Administration was established in Paris with seconded members from various other European Administrations. A member from the BPO joined this team and another took a post with the Commission. Following the issue of tenders, a network was implemented utilising the same technology as was used in the French Transpac system. Although not without its difficulties, it was successfully implemented and opened for service at the end of 1978 – see Figure 3. A network management centre was established in London. Traffic was quite low at the beginning because of Volume 5 Part 1 • 2011 | 57 the problems of interfacing the many hosts which were to provide the information content, a common problem that occurred every time a suitable specialised network was established. However, soon after its completion, it became obvious that the individual European Administrations were then actively implementing their own national networks and interconnecting them using X.75 links. Euronet was then in effect redundant as its traffic could be carried by these interconnected national networks. It was a difficult task for the CEPT to convince the CEC that Euronet should be closed down because it had been a success in that it had forced Administrations into providing their own packet-based networks even in those countries that had mainly gone down the circuit-switched path. Euronet was eventually closed down at the very end of 1984 [12]. Digitalisation of the trunk network As mentioned earlier, in 1976, I was promoted to deputy director and I moved to take charge of the planning and implementation within the trunk (long distance) network of an Integrated Digital Network (IDN). Perhaps it was significant that at the same time the CCITT special rapporteur for the packet mode of operation was also transferred to work on the proposed IDN in Norway. I recruited Keith Ward to assist me in my new role but, as mentioned, managed to retain my part time international responsibilities for Euronet. Studies had been completed into the feasibility of introducing digital switching and transmission into the UK trunk network in 1974 [13]. However the study was orientated towards integration of existing analogue switches, both Strowger and crossbar, as used in the transit network and the host of complicated signalling systems involved. After a quick analysis of the current situation it was decided that much of the feasibility report should be disregarded and that a start should be made almost from scratch on an overlaid digital switched network. Economic studies clearly showed that implementation was needed on a joint transmission and switching basis. Frank Thomas, as the then director of network planning department, endorsed this approach and, together 58 | Volume 5 Part 1 • 2011 Datapac Teleglobe Infoswitch Norway Sweden Norpac Swenet Canada Switzerland KDD Japan International data centre London Telenet Tymnet USA ITT RCA WUI NTT DS 50 International gateway Euronet Ireland Hong Kong and Singapore PSS UK national network Australia via OTC gateway France FRG Italy Denmark Netherlands Belgium Luxemburg Ireland UK Node transit international Transpac RETD Spain Existing Planned Figure 4: IPSS in 1978 with Geoff Simpson who headed the transmission divisions, the digital network began to take place. The key was that once a digital trunk exchange that switched 64kbit/s circuits was deployed, all transmission links to that exchange would be via digital systems with analogue to digital conversion occurring at the remote connected analogue exchange. Once further digital trunk exchanges were introduced, the links between digital exchanges would be fully digital and the analogue to digital conversion equipment would be moved to new locations. The first digital trunk exchange was opened in 1983 and by 1987 the trunk network in the UK had become virtually fully digital. My period planning for the digital trunk network was very rewarding because it was clear that digitalisation was feasible and necessary and thus there were little or no financial problems with actual implementation. It was also clear that, once an IDN had been established, the addition of local digital exchanges enabled the possibility of establishing an Integrated Services Digital Network which could provide the basis for a circuit-switched data service, the introduction of which (mentioned in Part 1 of this paper [1]) had been postponed in 1974. However following a major reorganisation in 1979 the network planning department as such disappeared and I became the deputy director (non-voice) located within the System X-orientated exchange systems department headed by John Tippler. My responsibilities covered not only data networks but telex and Datel services as well. The International Packet Switched Service (IPSS) The need for character terminal database access services to be extended internationally was recognised quite early and an international database access service to Tymnet via the three International Record Carriers in the USA was opened by the BPO using proprietary equipment and protocols in 1977. It quickly demonstrated the demand for international services. Following the definition of EPSS, public packet-switched networks had begun to be established in the USA, Canada, Spain, France and Japan, again based on national protocols. The new CCITT Recommendations were rapidly being adopted in these countries. Because of considerable international interest in data communications, the external telecommunications executive of the BPO (by then split into separate postal and telecommunications parts, the latter, later becoming British Telecommunications (BT)) decided that an international packet-based service was essential and IPSS was opened in 1978 [14] (Figure 4) with connections to the three USA International Record Carriers using Telenet TP4000 packet switches to CCITT Recommendation X.75 ratified in 1978. Connection to other countries followed also using X.75. As interconnection to EPSS was not technically feasible, access was via leased lines for terminals conforming to X.25 or via dial-up connections for terminals conforming to X.28. It was understood that once the BPO had introduced an inland X.25-compatible network, access would be via that THE JOURNAL OF THE INSTITUTE OF TELECOMMUNICATIONS PROFESSIONALS Opening date Glasgow Installed but not in operation at opening date Edinburgh 1981 extension Newcastle Figure 5: PSS equipment network. However, for various reasons that did not happen until 1981. Once opened, customers were transferred from the IPSS international gateways to the national nodes and the number and speed of international links were increased. Similarly the Euronet packet switches and network management centre, which were co-located with the IPSS packet switches, were subsumed into national/international services when Euronet closed in 1984. The international packet switching gateways, originally Telenet TP4000, were replaced by Alcatel Bell systems in 1986. With the IPSS, again charging issues arose and it was decided that IPSS would provide a virtual call packet service. This was a service option which ensured all packets followed the same path and it provided an easier way to implement the international accounting agreements between Administrations and/or Recognised Private Operating Agencies which historically had been introduced as a revenue sharing system for all international telecommunications services. The public PSS In 1976 I had been transferred to other duties and Brian Cross took over my UK responsibilities for data systems planning and had the daunting task of preparing the necessary papers seeking board approval for PSS. It had already become obvious that modifying the EPSS network to meet the new CCITT Recommendations was not a satisfactory solution and specifications for PSS were thus drawn up and finalised. Board approval was given in 1978 for its implementation. Eventually a contract was placed in 1979 with Plessey Controls (Poole) who had obtained a licence to utilise the Telenet TP 4000 Manchester Leeds Liverpool Birmingham Bristol Cambridge London Slough Reading IPSS NMC Figure 6: PSS at its opening in 1980 with the 1981 planned expansion packet switches. PSS was duly implemented and opened for service in the early part of 1980 [15] – see Figures 5 and 6. Access was by directly connected leased lines or dial-up via the PSTN, in both cases using modems from the Datel range. At a later stage, once digitally-based facilities such as BT’s KiloStream became available in 1984, these were also used for direct access links to the packet switching exchanges, better now known as nodes. PSS was of course connected to IPSS from the outset. Photo courtesy of Martin Baker. Elsewhere public data networks based on the packet mode of working began to make their appearance. One of the first was the “Red Especial de Transmisión de Datos” (RETD) [16] which was operational in Spain in 1971 offering terminal-to-host facilities as well as message switching. Telenet in the USA opened in 1975, Datapac in Canada in 1977, Transpac opened in France in 1978, Euronet in 1978 and others followed but PSS in the UK did not open until 1980. Early users of PSS were however the UK Banks. They had set up a Clearing Houses Automatic Payment System (CHAPS). At a given time each day they interchanged monies between themselves. To meet this requirement they were provided with a closed user group facility within PSS. However, to avoid problems with investors and the Stock Exchange if traffic flows were to become known in the public domain, dummy packets were transmitted at other times of the day. Thus a private switched network was established within a public network to the satisfaction of everyone. With the ordinary charging structure the amount of information recorded for a given transmission could far exceed the amount of data transmitted Volume 5 Part 1 • 2011 | 59 Telex and teletex Within the development department, at the same time as PSS was being developed, detailed specifications were being drafted for replacement digital telex exchanges to incorporate message switching as well as a teletex service, (not to be confused with the non-interactive broadcast teletext services associated with television transmissions). The teletex service had been fully defined by the CCITT SG I in June 1980 as an upgraded telex service which operated at 300bit/s. The service was strongly advocated by the Federal Republic of Germany, no doubt fully supported by their main contractor Siemens [17]. In the UK it was decided to go ahead with tenders for the supply of new stored program control telex exchanges with add-on options for a teletex and message switching service. A contract was eventually placed via GEC with their subsidiary Canadian Marconi Company [18] who in turn sub-contracted the necessary software to a Canadian firm who in turn subcontracted some software development to Norway. Hardware manufacture and installation proceeded on schedule but there were considerable delays with the software. Plessey Controls (Poole) were however given a separate contract to supply an exchange in Fleet in London complete References 1. Kelly, P.T.F. Computer communications – the early days 1966 – 1986 Part 1. The Journal of the Institute of Telecommunications Professionals, Vol. 4 Pt. 4, December 2010 2. Belton, R.C. and Smith, M.A. An Introduction to the British Post Office Experimental Packet-Switching Service (EPSS). Post Office Electrical Engineers' Journal, Vol. 66, p216, January 1974 3. Spooner, M.J., Wilson, E.J., and Neil, W. Experimental Packet-Switched Service: Procedures and Protocols Parts 1 and 2. Post Office Electrical Engineers' Journal, Vol. 67 p232, January 1975 and Vol. 68, p22 April1975 4. Després, R. RCP, The Experimental Packet Switched Data Transmission service of the French PTT. Proceedings of the 2nd International Conference on Computer Communications, Stockholm 1974, pp171-185 5. Roberts, L.A. and Wessler, B.D. The ARPA Network”. Computer Communications Networks, pp 485-500. Prentice Hall 1973 6. Barber, D.L.A. The European Computer Network Project. Proceedings of the 1st International Conference on Computer Communications, Washington 1972, pp192-200 7. Hadley, D. and Medcraft, D.W.F. Experimental Packet- 60 | Volume 5 Part 1 • 2011 Switched Data-Transmission Service. Network Design and Implementation. Post Office Electrical Engineers' Journal, Vol. 67, p88, July 1974 8. Eurodata European computer and communications market 1973-1985. PA Management Services S.A, Freiberg, Switzerland 9. The report of the National Committee on Computing Networks. BT Archives,1977 10.Davies, G.W.P., Gresser, J-Y, Kelly, P.T.F. and Thomas, J.R. The EURONET Telecommunications and Information Network. Proceedings of the 4th International Conference on Computer Communications, Kyoto 1978, p189 11.Kelly, P.T.F. The Euronet Telecommunications and Information Network. Institute of Electronic and Radio Engineers Journal, Vol. 49 No.11, November 1979 12.Broomfield, C.F. International Packet Switching Trends. Proceedings of the European Computer Communications Conference (Networks 86), London 1986 13.Breary, D. A long term Study of the United Kingdom Trunk Network. Post Office Electrical Engineers' Journal, Vol. 67, p37, April 1974 14.Ford, M.L. And Davies, F.W. International Data Networks UK Post Office Experience and Plans. Proceedings of the 4th International Conference on Computer Communications, Kyoto 1978, p41 15.Medcraft, D.W.F. PSS - The National Public Packet Switched Data Service. International Federation for Information Processing - UNESCO Comnet 81 pp 4-1 to 4-11 16.Alarcia, G. and Herrera.S. CTNE’s Packet Switching Network. Its Applications. Proceedings of the 2nd International Conference on Computer Communications, Stockholm 1974, pp163-170 17.Rüggeberg, R. The Development of the Teletex Service in the Federal Republic of Germany. Proceedings of the 6th International Conference on Computer Communications, London 1982, p338 18.Daniels, E.E. Telex Switching and Signalling. Post Office Electrical Engineers' Journal, Vol. 74 Pt. 3, p219, October 1981 19.Altehage, G. and Staudinger, W. The Public Switched Data Network of the Deutsche Bundespost: Three Years of Experience with Electronic Data Switching System (EDS) - Future Development of EDS. Proceedings of the 4th International Conference on Computer Communications, Kyoto 1978 THE JOURNAL OF THE INSTITUTE OF TELECOMMUNICATIONS PROFESSIONALS with a message switching element. Eventually the software problems in Canada were overcome, the detailed specifications having been relaxed to some extent, and the modernisation of the telex network went ahead and was completed in 1986. New generation telex machines were also introduced. The Fleet exchange never became operational and was in due course replaced by Canadian Marconi Company exchanges towards the end of the 1980s. The ad hoc options, namely teletex and message switching were never in fact introduced in the UK, probably due to the comments made in the NCCN report and in due course the availability of e-mail utilising the Internet. A teletex service was however introduced in some European countries, mainly those such as the Federal Republic of Germany, the Nordic group of countries and Austria who at that time saw teletex as part of a range of circuit-switched services [19]. The telex service was extremely popular and profitable during the 1980s. BT offered a prize to the person who became the 100,000th user of the service. But like some other services such as the Datel services it then went into decline and in the UK no further investment in the service was made. Modernisation of the telex network was in fact the last attempt to develop low speed circuitswitched data services in the UK but such services did not materialise. AUTHOR CONCLUSIONS Following experience with experimental networks like EPSS and with the finalisation of the X series of Recommendations, by the early 1980s X.25based packet-switched networks were extensively deployed and offered service worldwide. Not least of the battles that had been overcome were the charging principles and tariffs given the very different nature of the traffic compared to telephony. However, 64kbit/s circuit switching technologies leading to ISDNs and new high-speed digital transmission facilities leading to the Internet were emerging and saw the gradual demise of X.25-based packet networks. This stage of the story is addressed in Part 3 of this paper appearing in the next issue of The Journal. Acknowledgements I would specially like to thank Chris Broomfield for his contribution on International Services and also to all those who were associated with the introduction in the UK of the public data networks for their dedicated assistance and pursuit of objectives against often considerable objections and opposition. They should all be pleased with the outcome and the part they played at various times in its implementation. Abbreviations BPO BT CCITT CEC CEPT British Post Office British Telecommunications International Consultative Committee for Telegraphs and Telephones Commission of the European Communities Conference of European Postal and Telecommunications Administrations DCE Data Communications Equipment DTE Data Terminating Equipment EPSS Experimental Packet-Switched Service ICCC International Computer Communications Conference IDN Integrated Digital Network IPSS International Packet-Switched Service NCCN National Committee on Computer Networks PSS Packet-Switched Service Phillip Kelly About the author Phillip Kelly joined the GPO Engineering Department as a youth-in-training in 1944. He studied part time at various technical colleges, eventually graduating with an Engineering Honours degree in 1950. He then joined the Lines Branch of the Engineer in Chiefs Office. In early 1953, he was assigned as a systems engineer, to the Project Team set up to implement the first Transatlantic Telephone Cable System. Following the completion of the TAT1 Cable System, he moved to cover the planning and implementation of several European Cable systems including the Anglo-Swedish one. He then returned to the Oceanic Group to work on TAT 3. Awarded a Nuffield Travelling Fellowship in 1964, he had a year’s sabbatical study in Sweden and Canada. Returning to the UK he became involved with the design and implementation of computer systems for repetitive engineering tasks. He moved on secondment to HM Treasury in 1967 as head of the Governments Computer Division. When the Post Office was made a Public Corporation in late 1969, he returned to become head of the newly formed Data Systems Division where he was involved in complex studies and decisions regarding the use of packet or circuit switched systems for data transmission. He next became a Deputy Director of Network Planning overseeing the first plans for digital switching in the Trunk Network and then Deputy Director for non-voice networks. His final appointment was as Chief Engineer of Specialised Networks. In 1983 he retired on health grounds but soon afterwards became a consultant to BT Telconsult, and to manufacturers and IT companies. Later he took on the post of the first Executive Director of the European Electronic Mail Association a position which kept him involved until the late 1990s. He became a Freeman of the City of London in 1960. Volume 5 Part 1 • 2011 | 61