PRIVATE NETWORKING IN A MULTINATIONAL

Irene P. Harford

General Manager Office Automation Planning Massey Ferguson Limited Toronto, Ontario

Background

Massey-Ferguson is a Canadian based multi-national corporation that manufactures farm machinery and diesel engines.

 

It has 37 wholly owned factories in nine countries, and along with its associates and licensees, .Massey-Ferguson products are made in 85 factories in 31 countries.

Massey-Ferguson products are sold in 190 countries.

 

With this type of international company structure, and a Head Office based in Toronto, it is essential that information flows quickly and easily among international Operating Units and the Head Office function for all financial information and reporting statistics.

 

In the sixties, the fastest common communications medium internationally was telex/ teletype, and large telex centres evolved, rather than being designed, at main communication centres in North America - such as Des Moines, and in Coventry, England, which was the main communications centre for Europe.

 

Manual telex refile, using torn tape techniques was the method used to disseminate messages to other Massey-Ferguson and Perkins locations in North America, Europe, and the Southern Hemisphere. Point to point leased lines were installed as traffic costs justified the tariffs involved. Early message switching facilities were rented from the British Post Office (B.P.O.) in the U.K. at London, England, and from Western Union in New York when we cost justified our first momentous quarter speed telegraph circuit between the U.K. and North America.

Network Beginnings

Telex volumes increased rapidly, and the international trunk link of our growing network was increased to half speed in 1973.

However, as our traffic volumes grew, the technical differences of our switching facilities became more apparent. Significant compatibility problems existed between the Philips System offered by the B.P.O. and the Western Union switch. Both systems were progressively upgraded independently but many of the features offered were excluded from our use because of incompatibility in working the systems together.

 

It is significant to mention at this juncture that at this time the telecommunications function was fragmented within our company structure and did not have a single staff reporting route to a Head Office function. This complicated finding an effective method of reporting problems and failures of our first stage international network, quite simply because it was not identified as a single area of responsibility within our company structure.

 

A General Manager of Telecommunications was appointed in 1974 reporting to the Corporate Management Systems Director at Head Office and his first task was to produce a solution to the problems of our apparently ever expanding telex traffic demands and inadequate methods of routing and delivering telex messages throughout the network. A consolidated measurement of traffic volumes produced some startling figures.

 

The cost justification of the use of private message switching facilities in the U.K. became a relatively simple management decision, and similar rented facilities with emulated software were arranged with CNCP in Toronto, Canada. Other Operating Units internationally were added to this network from both the U.K. and Toronto message switches.

 

A new switching centre was set up in the Coventry area of the U.K. at Stoneleigh in Warwickshire using the ITT ADX 600 for all international port terminations in Europe.

 

Principally for company domestic and structural reasons, a further ADX 600 was installed at Perkins, Peterborough in the U.K.

Introduction to Time-Sharing

In 1975, Massey-Ferguson introduced I.P. Sharp's APL time-sharing facilities for financial information reporting systems in Europe and North America, initially on a dial-up basis.

 

APL became the principal financial information reporting system and a means of producing answers to those hypothetical "what if questions which have to be answered in any corporation.

Time-sharing occupancy times soared rapidly.

 

In May 1978, processing was transferred from I.P. Sharp's mainframe in Toronto to Massey's Data Centre in Toronto. Concentrator units of the same type as used by Sharp on their own network, the Alpha 16, were installed in Stoneleigh and Peterborough, England; Hanover, Germany; Lucerne, Switzerland; Le Plessis Robinson, Paris, France; Aprilia, Italy; and in Toronto.

 

Back-up facilities from Sharp were maintained with a link from the Toronto Alpha to Sharp's Alpha in Toronto.

 

Telegraph, as an ASCII based system with nominal transmission speeds of 50-110 baud was operationally linked with the 300 baud asynchronous APL via the ubiquitous circuit we termed "Circuit X". This circuit theoretically made every telex terminal a time-sharing terminal (if you could put up with the enormous response delay) but more importantly made every APL terminal capable of sending a telex interfacing to the private message switching service.

 

The trunk carrier circuit between East and West was multiplexed to provide 4800 bps carrying time-sharing service, and two further channels at 2400 carrying IMS and MVS, using Racal Milgo Multimode 9600 modems at each termination to provide the separation in operating bandwidth.

Rationalisation of Data Centres

Up until this time Massey-Ferguson had been operating its Data Centre Operations on a relatively autonomous basis, with IBM 370/158's in Des Moines, Toronto, Coventry, Peterborough, Hanover, Paris; a 370/148 in Rome and an Amdahl V5 in Aprilia. Italy; a 370/138 in Canton, Ohio, and a 370/145-500 in Sao Paulo.

 

A rationalisation of Data Centres, which has significantly increased the operating flexibility and disciplines of a multinational data network, has resulted in two main Data Centre Utilities; one serving North American operations, located in Des Moines, Iowa, and the other serving European locations (with the exception of Perkins, Peterborough) based at Birmingham in the U.K. using an Amdahl V7 in Des Moines and an Amdahl V8 in Birmingham.

 

The ITT Comten box is used to provide the interface to the mainframe running IMS/MVS at the various locations.

 

Conversion from local 158's to main Data Centre Utilities is due to be completed by the end of the year which brings us up to date in the evolution of our international data network, which has progressed in 10 years from a very low speed data network to a major private international network with operating speeds of 9600 over two international bearer circuits.

National Voice Network Additions

During the past few years, however, we have also developed and installed two tandem switched voice networks — one in the U.K. and the other in North America.

In the U.K using Coventry as the tandem switching node, we are using four widebands, i.e., 4 x 48 KHZ, each of which is multiplexed to provide 12 channels for both voice and data; and a channel which can be engineered to further multiplex 10X telegraph channels.

The network is the first private stored programme controlled voice network in the U.K The terminal PBX's and tandem switch are Plessey PDX which is made under licence in the U.K from the original ROLM CBX. This is capable of handling local data transmission switching in addition to voice.

The tandem is capable of handling 96 ports, and is at the moment being used for national voice circuits only. Obviously, we wish to include international circuits for voice transmission operating on an auto/auto basis, which is currently constrained because of the problems associated with international agreement on signalling between Europe and North America on private circuits.

 

We have cost justification to install at least one circuit to Toronto and others to Germany, France and Italy.

 

The North American voice network is now installed and running using a switch in Detroit; a star network design with some 13 location terminations.

 

It is planned to link these two networks together. However, international leased lines for voice only transmission, especially when examined as an analogue transmission and the bandwidth required, are extremely expensive on long haul routes, and also give an extremely poor utilisation of bandwidth availability, particularly when measuring the small window of time available during normal working hours between East and West.

 

It is with this in mind we have been carrying out trials on a speech digitiser box which would allow us to transmit slightly impaired speech transmission over a 2400 bit channel of an international circuit.

Future Network Rationalisation

Our objective, must be to merge all networking applications regardless of type.

This becomes a more obvious essential criteria as distributed processing techniques become more readily acceptable and understood.

 

Our company lays great emphasis on the development of information processing techniques throughout the organisation, which will provide us in the following years with a single electronic mail system, interfacing to the public networks, as and when they become available; electronic storage; access to and from both private and public data bases; picture as well as text transmission and all systems integration which is generally categorised as office automation.

 

The future phases in the rationalisation of our private international network must consolidate the various stepping stones of switching/concentrator boxes such as the ADX's, the Eclipse, Alpha 16, and Comten, into ideally one such box which is intelligent enough and flexible enough to accept and evaluate various protocols and control functions; has adequate port and bandwidth capability and will accept varying transmission speeds and store and forward.

Similarly we are evaluating the use of terminals. The proliferation of terminals over the past five years is horrendous as terminals have been developed essentially on the basis of one box: one application. In consolidating our systems, this concept becomes blatantly outdated and ludicrous.

 

We consider it inevitable that there will be many steps to take in reaching our objectives, but we are currently looking at various word processing terminals which have the potential of being used as a common type of terminal and at least initially cover several applications in addition to its basic use as a very smart typewriter.

 

The potential opportunities operationally and financially are challenging, exciting and without question offer enormous rewards. The next 10 years hold a great deal of promise for a new dimension in networking operations.

 

 

 

EDITOR’S NOTE

The following two diagrams show the evolution of the Massey network before the introduction of packet switching technology.

 

 

EDITOR’S NOTE

The Alpha 16 boxes in the next diagram are IPSANET nodes. The CN Eclipse and ADX boxes are Telex switches. The subsequent diagram shows a plan to install NCR COMTEN 3805 processors sold by ITT. These were somewhat more powerful than the typical IPSANET node as you can see from the parallel 9600 bps links connecting the Birmingham FEP with remote processors.