Chapter 13 - Data Communications: Adaptation 1979-1986
13.11 Tymnet and the Caravan Project 1982
In early 1982, Sarah Schlinger, director of business planning of Tymnet, decided she had learned enough. It was time to prepare a business plan and present it to her superiors. If her responsibility was to articulate how Tymnet might more aggressively pursue the coming opportunities in wideband communication services, she had potentially found a jewel. Her only hesitation: the opportunity was to build a product for resale, not a service to sell more time on Tymnet’s packet-switching network or private networks to corporate customers. But the fact that the idea had emerged from an on-going project Tymnet had with Satellite Business Systems (SBS) gave her confidence the idea to build a T-1 multiplexer would not be dismissed out-of-hand.
Tymnet had been created around 1979 to administrate and develop the packet-switching network Tymshare had spent a decade building to serve their time-sharing customers. Tymshare recognized corporate customers needed to interconnect their growing base of disparate computers and peripherals - a need independent of the need for time-sharing services – and through Tymnet they could provide such networks either as a service or by creating private networks to then sell. The project that had led to the idea of building a T-1 multiplexer came from an experiment Tymnet was conducting with SBS that would interconnect a customer’s computers/peripherals in New York City and San Francisco using Manhattan Cable in New York City, SBS’s long-haul satellite connection and microwave radio services in the Bay Area. Audrey MacLean, Western Regional Manger of Tymnet, and instrumental in selling a $6 million private network to the Bank of America, remembers:
I had responsibility for helping to come up with some willing guinea pigs, customers, for that experiment; among them folks like Crocker Bank, etc. However, my interest in the area was keen, because at the time, the western region was really leading Tymnet into a business area that for them was private packet-switched sales. It was probably in ‘81 that Tymnet first started building private packet-switched networks, beginning with TRW down in southern California, and then, in the 1982 time frame, the Bank of America network. I looked at what the Bank of America’s overall communications need was, and saw that the X-25 network that I had sold them really only covered only a small cross-section of their application needs, and that if I looked at their whole global data network plan, and if I also concurrently looked at their voice network that they were rolling out at the time – I think they bought $15 million of PBXs at about the same time they bought the $6 million worth of packet switches – it began to occur to me that it would make logical sense for them to, in some fashion, at least in terms of the transmission, integrate these two networks.
The experiment was looking at exactly that, provisioning T-1 rate service out to the end-user’s location, T-1 being a 1.544 megabit transmission connection to that location, which by far exceeded any of the then current demands for data connectivity, fifty-six kilobit was considered really fast, but was entirely appropriate if you were going to look at including voice switching needs, or voice traffic needs, on those links.
The challenge was how to interconnect and switch the multiple T-1 satellite circuits and the multiple data and voice circuits (or DS-0 circuits) at customer’s locations. MacLean remembers:
There were multiplexers, such as the type made by General DataComm, that could take N inputs and one T1 output. There were channel banks that could take 24 or 48 voice inputs and one T1 output, but they were all the traditional multiplexer triangles, if you will, in terms of their architecture. They were not switches. So in order to put together a switching core in these main hub locations in the network, you literally had to back-to-back cross wire multiplexers to begin to have switching.
Roger Chrisman, who had joined Tymnet to head up the engineering for the SBS project and reported to Art Caisse, had the responsibility to create the product to interconnect and switch the T-1 or DS-1 circuits. MacLean remembers:
Roger used to walk around disgruntled, bitching about the fact that this was one step up from carrier pigeon. So, it was as his thinking began to grow about the kind of switch that he needed to solve the problem that he came up with the thinking of building a switch that handled multiple DS-1s and was able to move traffic from one DS-1 onto another within the same switching mechanism.
This was the product idea Schlinger had studied and reduced to a business proposal and submitted to her Tymnet superiors. They in turn took it to the management of Tymshare. Again MacLean remembers:
A proposal was taken forward to the board of Tymshare, the parent company, to build the kind of switching product that Roger was lamenting didn’t exist for his experiment, and the board of Tymshare turned it down, primarily because they didn’t envision themselves as a hardware company. They saw themselves more as a user of other people’s hardware, and provider of X-25 software technology.
MacLean, Chrisman, Caisse and Schlinger were both disappointed and partly relieved. If Tymnet did not want to act, then their consciences were clean if they did.