From mainframe to laptop

A history of computing at Yale

By Kushal Dave

KOI ANUNTA/YH

A student might be inclined to look at the present state of Yale computing with slight disappointment — Ethernet connections can be difficult to get working, Computing Assistants(CAs) can be hard to contact, computers from Yale's MicroComputer Support Center are sometimes misconfigured, and the school webpage has a rather weak design — never realizing that Yale was once a hotbed of computing innovation. Even today, Yale is doing a pragmatic job of keeping up.

The center of this activity is 175 Whitney Avenue, home to Information Technology Services (ITS). Once, this building housed Yale's early mainframe computers, massive room-sized machines that did a fraction of the the computing that a typical desktop does today. Along with the rest of the world, computers at Yale have come a long way.

In the beginning...

Computers have been around since the 1960s, when the University was home to a prehistoric International Business Machines (IBM) 709 mainframe. This mainframe was a central computer that all users worked from. It was located in the basement of what is now the Yale School of Management and was used by the astronomy department. In 1965, Yale purchased what was then the top-of-the-line computing machine from the DCS corporation. A short while later, faster and more poweful mainframes were released, antiquating the DCS machine, and the university made its first (though certainly not last), mistake in hardware purchasing. The new mainframes had the ability to connect to terminals in other locations, allowing remote access. The DCS, however, was used until 1971 for academic purposes. Of course, in the 1960s, new computer models were not released as frequently as they are now — large computer purchases were intended to last. "That was back when you could conceive of a computer with a lifespan of eight years," Howard Gilbert, Senior Research Programmer of Technology & Planning, of ITS said.

'Yale was in the vanguard.'

During the '70s and '80s, Yale assumed a role of leadership in the area of software for the new mainframes. In the 70s a school could make money by developing software and then selling it — the price of a program ran in the tens of thousands of dollars at that time.

Yale computer specialists wrote software that addressed specific problems that came up on a daily basis. One piece of Yale-authored software, the Conversion Yale Terminal Operating System (CYTOS), allowed time sharing on mainframes. While IBM machines were initially designed for batch processing, where one user has complete control of the system, time-sharing allows users to work concurrently.

Philip Long, head of Academic Computing Services (ACS) explains this was a revolutionary step. "It was one of the first time-sharing systems in the world," Long said, explaining that it was used for several years until IBM released its own system. "Yale was in the vanguard."

Another innovation from Yale's computing gurus was establishing a connection between IBM terminals and terminals created by DEC and other makers that worked with ASCII (American Standard Code for Information Interchange). Though the DEC and IBM terminals were the interface to the same set of mainframe computers, they operated on two different protocols. According to Gilbert, Yale became the "industry-known leader in bridging the gap," between the two protocols. At one point, even the White House and IBM both used Yale ASCII) software. Even now, some of the Yale applications from the '70s are still in use.

Yale was also the birthplace of an Internet precursor. Starting with an initial 9,600 baud connection between City University of New York (CUNY) and Yale around 1979, a network called "Because it's there Net," (Bitnet) was established. Bitnet provided a link for professors to communicate with one another via email, though oftentimes the subject of these email conversations would be computers themselves. Bitnet rapidly expanded, until it eventually spanned the globe, connecting thousands of university computers. Anybody could join on as long as they paid for the connection between themselves and the rest of the network. Although the Internet soon surpassed Bitnet because of its capacity to go beyond simple email and government funding, Bitnet was an impressive system in its time.

The PCs arrive

In the '80s, the introduction of personal computers brought about massive change, after an initial period where "the first generation of PC's, both from IBM and Apple, could be conceived of as toys," Gilbert said. "You could see they would eventually be good for something."

As PCs first emerged, Yale's computer scientists worked to make them resemble mainframe terminals so that PCs could be interfaced with older systems. It was clear, though, that PCs would soon be able to run programs off a seperate operating system. Long pointed out that this was the first step in decentralizing computing. "The computer began to move out of the office and into the room, and you didn't have to be a department to have a computer," Long said. With a major grant from IBM and assistance from other vendors, Yale began to place computers in residential colleges and Cross Campus Library. Also in the '80s, the first PC networks were created using coaxial cables, the same technology used to carry cable television. Ethernet adapters cost a few thousand dollars at that time, and so the now vast Ethernet was limited to the interior of ITS. In addition, there was an AppleTalk network, which provided a cheap, though inefficient alternative, as the AppleTalk network was not compatible with the many IBMs Yale owned. The old coaxial cables were laterreplaced with the 10baseT wiring that runs Yale's fast Ethernet and large Local Area Network.

While technology marched on, ITS experienced a few organizational changes. At one point, Computer Information Systems (CIS), now called ACS, the Administrative Shop (ADS), Data Network Operations (DNO) and the Computer Science Department all operated independently. ACS went through many incarnations as well, such as the Yale Computing Center, as which they reported to the president.

A reorganization of the old structure gave birth to ITS's role as an umbrella organization for all other computer-related groups. According to Gilbert, the main characteristic of ITS at present is decentralization, "rebelling against totalitarianism." There is not even a unified budget, serving instead as an "umbrella of budgetary units," he said. However, Gilbert confesses that there are "rumors" of a move to greater centralization.

"The important thing to realize is that in this scheme, ITS has no strong leadership role," Gilbert said. "The Yale administration does not appreciate anyone proselytizing." In the current system, the impetus is on the faculty to articulate what they need. However, Gilbert explains that faculty do not always know what will meet their needs. He describes the lack of a corporate chief information officer to control everything as a "structural idiosyncrasy." This problem may never be solved, since some ITS staffers still feel that it is impossible to settle upon a good, lasting structure in an industry that constantly changes.

"I think it would be inaccurate to say Yale or any other major university knows what the hell its doing in terms of information technology," Gilbert said. "Reorganization can be defeating if it is done too often." Thus, while the ITS structure is not perfect, it works.

The central question

As ITS looks to the future, and tries to negotiate some of the obstacles in its way, such as the balancing act between the Macintosh and PC platforms, issues of security, and the unpredictable computer industry, Yale's computer policy remains fairly conservative.

"The Yale professional staff has tried to provide the highest quality service available within the context that it isn't the top-of-the-line of anything, and we've been quiet successful in that regard," Gilbert said. He points out how at Brown, they installed coaxial cable at high expense throughout the campus, only to have it become antiquated when 10baseT Ethernet arrived.

"By not aspiring to be the first, but waiting until the technology became mature, became cost-effective, we've done a pretty good job of delivering high-quality service in the mainstream. We've a had a certain amount of respect in that regard."

Long also acknowledges the difficulty in setting policy. To him, the central question is, "how do you choose what is really effective and meaningful?" He explains, "We can add two and two much faster than we used to, but that doesn't make an information age." After all, not all change is for the better, though one clear benefit from contemporary computing is its capacity for "connecting people with people."

What do you think? Respond in Speak your Mind.

Back to Online Features...