‘Mainframe Renaissance’ Is Really the Revenge of the Non-Techies

Is the mainframe computer back in style?

It seems a ridiculous question to ask in this day and age. IBM, the leader in mainframe computing forever, has seen its sales of mainframes cut nearly in half over the past five years. Sales of its minicomputers, in which it is also the market leader, declined for the first time last year by about 10%. Other mainframe or minicomputer companies are either dying or stagnant.

Nonetheless, the corporate keepers of computing argue that mainframes are roaring back. My colleague at Computerworld, Editor-in-Chief Bill Laberis, wrote an editorial a few weeks ago titled “Big Box Redux.”

“It seems that after suffering through the dark ages of hyperbole, overselling, misrepresentation and outright fraud,” he wrote, “users are now being sung a different tune by vendors, analysts and others about client-server computing and the real world.”

Indeed, you can observe--at gatherings of people who manage corporate computer systems--a kind of growing ebullience, a sense that the world has come to its senses and will let the computer managers once again do their jobs as they should.

The fortunes of the mainframe computer are a kind of analog for the balance of power between the keepers of computing inside large companies and the users of computing--executives, owners, managers and workers.

For years, between about 1965 and 1985, the non-technical people in corporations were held hostage by the technical people. Mainframes and minicomputers were so expensive, so difficult to comprehend, so unrelated to the skills of managing the functional areas of a company, that the people who knew how to run them could affect the most basic fortunes of the company. And the functional managers weren’t able to tell how they did it.

The classic case of this occurred when American Airlines’ computer people figured out how to make all the reservations on their computers, thereby taking control of an entire industry’s ability to record its transactions with customers. American’s “Sabre” system has been cited repeatedly as an example of how information was becoming a functional area itself, rather than just a service for the other functions in the company. It was enough to make the non-computer people feel like the computer people were taking over the business.

So this question of the status of the mainframe can be pretty important. The power of computing officials has been declining along with their big, centralized computers. In job surveys, “chief information officer” ranks as the most unstable position in executive ranks.

It’s no wonder then that after more than a decade of being roughed up by the chaos of millions of personal computers, computer managers want to believe they are getting back in control.

The change that lets them believe this is the arrival of “client-server computing,” which has transmogrified from a description of a particular kind of computing into a kind of clarion call for what Computerworld’s Laberis calls “a more structured mainframe mentality.” Others might describe it as a call for returning to Soviet-style, centralized control of information processing.

In the old days, mainframe computers were controlled by terminals attached to the central processing unit by wires. The rise of personal computers represented a real problem for corporate computing managers, because PCs weren’t attached to anything and could be controlled by individuals.

Then groups in corporate departments started linking PCs together to keep track of small pieces of information that hadn’t been important enough to manage on mainframes or minicomputers. The groups started to link each other together into loose networks of networks, often for the purpose of using electronic mail.

By last year, most big companies had launched some kind of formal effort to begin to integrate all of these networks and all of their mainframe systems into a single integrated system.

Enter client-server computing.

The original idea of client-server computing was to describe the notion that, in a larger system, you would have large and small computers cooperating with each other, as “clients” requesting services and “servers” providing services.

The idea was to differentiate this kind of computing from the old kind of computing--described as “host-terminal” or, more appropriately, master-slave. Servers and clients are supposed to cooperate and spread the computing around the system in a perfect digital representation of a kind of free-enterprise economy.

In their attempt to regain some semblance of control over the fortunes of the company at large, the corporate computing officials are now looking at client-server computing as their salvation. They see a vast sea of wires connecting together all those obstreperous personal computers--and they believe they can once again gain control of the computing resources of the company.

What they fail to recognize is that client-server computing is actually the ultimate revenge of the non-technical managers.

All the economic pressures are for vendors to build small, inexpensive components that can be plugged into these vast information systems by anybody--even those without serious technical training. The portion of the system that will be managed by the computing officials will inevitably become a smaller piece of the whole pie. And the functional managers of the company will gain increasing control over the resources they need--whether to design and manufacture products, to market and distribute them or to manage the operations of the company.

The mainframe may look like it’s back in style. But it won’t be long before it becomes clear that 1994 was just the last gasp of a dying technology.