Packet Switch Stream

In the United Kingdom, Packet Switch Stream (PSS) was an X.25-based packet-switched network, provided by the British Post Office Telecommunications and then British Telecommunications starting in 1980. After a period of pre-operational testing with customers (mainly UK universities and computer manufacturers at this early phase) the service was launched as a commercial service on 20 August 1981. The experimental predecessor network (EPSS) formally closed down on 31 July 1981 after all the existing connections had been moved to PSS.

Description

Companies and individual users could connect into the PSS network using the full X.25 interface, via a dedicated four-wire telephone circuit using a PSS analog modem and later on, when problems of 10-100 ms transmission failures with the PCM Voice based transmission equipment used by the early Kilostream service were resolved, via a Kilostream digital access circuit (actually a baseband modem). In this early 1980s era installation lead times for suitable 4-wire analog lines could be more than 6 months in the UK.

Companies and individual users could also connect into the PSS network using a basic non-error correcting RS232/V.24 asynchronous character based interface via an X.3/X.28/X.29 PAD (Packet Assembler/Disassembler) service oriented to the then prevalent dumb terminal market place. The PAD service could be connected to via a dedicated four-wire telephone circuit using a PSS analog modem and later on via a Kilostream digital access circuit. However most customers, for cost reasons, chose to dial up via an analog modem over the then UK analog telephony network to their nearest public PAD, via published phone numbers, using an ID/password provided as a subscription service.

The current day analogy of ISPs offering broadband always on and dial up services to the internet applies here. Some customers connected to the PSS network via the X.25 service and bought their own PADs. PSS was one of the first telecommunications networks in the UK to be fully liberalised in that customers could connect their own equipment to the network. This was before privatisation and the creation of British Telecommunications plc (BT) in 1984.

Connectivity to databases and mainframe systems

PSS could be used to connect to a variety of online databases and mainframe systems. Of particular note was the use of PSS for the first networked Clearing House Automated Payment System (CHAPS). This was a network system used to transfer all payments over £10,000 GBP (in early 1980s monetary value) between the major UK banks and other major financial institutions based in the UK. It replaced a paper based system that operated in the City of London using electrical vehicles similar to milk floats. Logica (now LogicaCMG) designed the CHAPS system and incorporated an encryption system able to cope with HDLC bit stuffing on X.25 links.

Speeds

There was a choice of different speeds of PSS lines; the faster the line the more expensive it cost to rent it. The highest and lowest speed lines were provided by the Megastream and Kilostream services, 2M (Mega) bit/s and 256K (kilo) bit/s respectively. On analog links 2400 bit/s, 4800 bit/s, 9600 bit/s and 48 kbit/s were offered. Individual users could link into PSS, on a pay as you go basis, by using a 110, 300, 1200/75, 1,200 or 2,400 bit/s PSTN modem to connect a Data Terminal Equipment terminal into a local PSS exchange. Note: in those days 2,400 bit/s modems were quite rare; 1,200 bit/s was the usual speed in the 1980s, although 110 and 300 bit/s modems were not uncommon.

History of implementation

The International Packet Switch Stream (IPSS) is an international X.25 network service launched by the international division of BT to which PSS was linked to other packet switched networks around the world. This started in about 1978 before PSS went into operation due to the high demand for affordable access to US based database and other network services. A PAD service was provided by IPSS to this market in advance of PSS launch.

For a brief time the EEC operated a packet switched network, Euronet, and a related project Diane to encourage more database and network services to develop in Europe. These connections moved over to PSS and other European networks as commercial X.25 services launched.

Later on the InterStream gateway between the Telex network and PSS was introduced based on a low speed PAD interface.

The network was initially based upon a dedicated modular packet switch using DCC's TP 4000 communication processor hardware. The operating system and the packet switching software was developed by Telenet (later on GTE Telenet). At the time of PSS's launch this was in advance of both Telenet's own network and most others that used general purpose mini-computers as packet switches.

BT bought Telenet's system via Plessey Controls of Poole, Dorset who also sold Telex and Traffic light systems.

Later on BT used Telematics packet switches for the Vascom network to support the Prestel service and also bought the Tymnet network from McDonneld Douglas.

In the words of BT's own history:

British Telecom purchased the Tymnet network systems business and its associated applications activities from the McDonnell Douglas Corporation on 19 November (1989) for $355 million. Its activities included TYMNET, the public network business, plus its associates private and hybrid (mixed public and private) network activities, the OnTyme electronic mail service, the Card Service processing business, and EDI*Net, the US market leader in electronic data interchange.

BT Tymnet anticipated developing an end to end managed network service for multi-national customers, and developing dedicated or hybrid networks that embraced major trading areas. Customers would be able to enjoy one-stop-shopping for global data networks, and a portfolio of products designed for a global market place.

These services were subsequently offered by BT Global Network Services, and subsequently by Concert as part of Concert Global Network Services after the Concert joint venture company was launched on 15 June 1994.

Exchange for other assets

It is believed BT subsequently exchanged major US elements of the Tymnet business with MCI for other assets when the proposed merger of their two businesses was thwarted by MCI's purchase by WorldCom.

The last PSS node in the UK was finally switched off Wednesday, June 28, 2006.

Network management had been run on a system of 24 Prime 63xx and 48xx computers running a modified versions of Revisions 20 and 22 of the Primos operating system. These network management systems were based in London and Manchester. Packet switches were installed at major trunk exchanges in most major conurbations in the UK.

The DNICs used by IPSS and PSS were 2341 and 2342 respectively.

BT's attitude to packet switching was ambivalent at best. Compared to France's Transpac that had a separate commercial company with dedicated management and saw X.25 packet switching as a core offering BT's then senior management regarded packet switching as a passing phase until the telecommunications nirvana of ISDN's 64 kbit/s for everyone arrived.

Even in its recent history BT's senior management stated that the Internet was "not fit for purpose".

Investment challenges

PSS suffered from inconsistent investment during its early years. Sometimes not enough and sometimes too much but mostly for the wrong reasons.

Investments in value added network services (VANS) and BT's own access level packet switching hardware delayed operating profit. This in turn dented PSS's low credibility with BT's management still further. Despite healthy demand for basic X.25 services and the obvious trend for more demanding bandwidth intensive applications that required investment in more powerful switches a decision to develop BT's own hardware and network applications was made instead.

In the midst of this IBM (the then market leader in computing) and BT attempted to launch a joint venture, called Jove, for managed SNA services in the UK. And for a time significant extra expenditure was allowed for BT's data services, PSS being the major part, as one concern of regulators was this joint venture might damage work on Open Systems Interconnection. This only made cost control worse and achieving operating profit delayed further. Eventually the UK government decided the SNA joint venture was anti-competitive and vetoed it. But not before PSS management was allowed to commit to large investments that caused serious problems later.

One of the few successful value added applications was the transaction phone used to check credit cards by retailer to validate transactions and prevent fraud. It was believed that putting a packet switch in every local telephone exchange would allow this and other low bandwidth applications to drive revenue. The lesson of Tymnet's similar transaction phone that just used a dial up link to a standard PAD based service was not followed. Each low end packet switch installed added costs for floor space, power, etc. without any significant value added revenue benefit resulting. Nor were they adequate for X.25 host traffic.

Ideas like providing a more user friendly menu based interface, called Epad, than X.28 was proven obsolete by the advent windows based clients on PCs.

As the added value services, named PSS Plus collectively, added significant costs and headcount while contributed virtually no revenue a change in PSS's management eventually resulted. While a decision was eventually made to put some of the basic network services people in senior positions and try to launch what had been developed this proved to be a major mistake. An exodus of people who were developing the value added network services helped reduce some costs. However significant on-going expenditure had been committed already to manufacture packet switch hardware and by using the very expensive Tandem computers in existing VANS. Operating profit was still not achieved and a further change in management with McKinsey consulting being called in.

McKinsey's recommendation that increasing revenue while cutting costs was required to turn around the business was duly followed by the new management and an operating profit achieved in about 1988. This rested on running PSS efficiently and cutting the VANS as much as possible. PSS was then merged with other failing business like Prestel as it became part of a larger Managed Network Services division that was used to fix or close BT's problem businesses.

Later history

While PSS eventually went the way of all X.25 networks and was overwhelmed by the internet and more significantly the internet's superior application suite and cost model. BT did not capitalise as much as other packet switch operators by subsequent mistakes concerning the internet, Tymnet, BT's North American operations and the Concert Global Services with ATT.

BT's failure to become the major ISP in its own home market unlike every other former PTT and the success of Dixon's Freeserve, Demon and Energis based virtual ISPs in the same sector has only been recovered from recently. Only after BT changed its most senior management who were fixated on circuit switching/ISDN based on System X/Y telephone exchanges and embracing broadband/internet lock stock and barrel has this changed. An emergency rights issue also helped resolve the debt from acquiring second or third ranked old telcos style companies around the world.

Now BT appears to be inheriting a dominating position in the Global Network Services market, based on packet switching, as CSC and Reuters sell up their networks to BT. As the commodity price of IP services based in their core 21st century MPLS network to carry voice and data finally gives them the real cost efficiencies that packet switching always promised.

gollark: Most rural areas, like where I am, only have VDSL lines from BT.
gollark: > do people have these problems in other countries?In the UK, we have *different* problems.
gollark: Meh, I turn that off anyway because it seems to be more "Microsoft-approved boot" than *secure* boot. Unless you bother to set your own keys, I guess.
gollark: No, I did, several times.> The data/body can be large, contain arbitrary bytes, and is actually meant to store large amounts of data.> - servers may allocate limited-sized buffers for incoming request headers so you can't put too much in them (this is somewhat problematic for cookies)> request bodies can probably be handled more performantly because of stuff like the length field on them> - request bodies are generated by forms and all sane clients so stuff is mostly designed to deal with those
gollark: I did say multiple times why that's not really a good idea.

See also

This article is issued from Wikipedia. The text is licensed under Creative Commons - Attribution - Sharealike. Additional terms may apply for the media files.