Client/server religion: should you convert?

In the beginning, there was the mainframe. On the second day, it begat a theology in which proselytes worshipped at the altar of the giant black box, and high priests of IBM and other technology companies carried glad tidings with missionary zeal.

Before resting on the seventh day, the high-tech deity created the client/server configuration, complete with a radical new value system based on computing speed and low cost. Like the early followers of Martin Luther, converts rejoice at their liberation from orthodoxy and the ability to make more decisions themselves.

Like religion, however, client/server can be misunderstood and misused. The technology isn't necessarily cheaper, and, depending on your current system, it isn't necessarily better.

In considering a changeover, mull none but these two corollary commandments: Honor thy business process, and never place technology idols before it.

GAUGING THE GROUNDSWELL 

The client/server movement initially was driven by vendors quick to spot a marketing opportunity. Here's how it works: A "client" computer, usually a personal computer or workstation, is connected to a "server" or back-end system. The idea is to split the work between the client and server machines, thereby balancing the processing load and giving system designers greater flexibility in program development and data storage.

Customer conversions soon formed a groundswell. In theory, client/server would reduce capital expenses and virtually eliminate "big iron" mainframes. That appealed to cost-conscious companies in the thick of a recession.

Unfortunately, some installations haven't lived up to expectations. In fact, they threw into doubt many things users of bullet-proof mainframes took for granted, including reliability and data integrity.

One hotel chain spent a bundle on a client/server system only to find one of the applications it runs can bring the server to a standstill. A northern California maker of window coverings lost nearly $5 million when its much-heralded client/server system crashed and took the entire accounts receivable data base with it.

No one explained to these companies they would have to reinvent their infrastructure and applications. Or that they would have to train or retrain people to install, program, and manage these machines.

The moral of the story: If you think client/server is going to save you money, it won't--at least not in the short term. In fact, one consultant has cautioned companies not to expect significant savings from client/server projects for at least five years.

DANCING WITH DINOSAURS

By contrast, you've been running your business on mainframes for years. They're not terribly expensive to maintain, you've capitalized and depreciated the equipment, and you have hundreds of IS staffers who have spent millions of man-hours learning how to use them. In fact, some applications are ideally suited for mainframes, including number-crunching programs that don't require much attention from end users.

Thus, some natural questions--and answers:

* Should you replace your mainframes with client/server systems? No.

* Should you move to client/server? Yes.

* Should you keep your mainframes? Yes.

Contradictory? Not at all. If you are focusing on client/server technology, then you are not following the true tenants of the religion. Business decisions must dictate technology decisions, not the other way around. CEOs should worry about profit margins, cycle times, and market shares--not about whether to adopt new technology.

If you decide to implement client/server because you think it's going to cut costs, or because your competition is doing it, stop and reconsider. You should implement technology only because your business processes either function, or have been re-engineered to function, in a way that will take advantage of client-server.

THE GOSPEL ACCORDING TO ST. PARK

The good news is you can move to the new technology gradually and sensibly. Doing it step-by-step will get you more productive workers and greater operating latitude.

For example, rather than "downsizing" your most critical business applications, you can build new ones. Or, you can migrate older, "legacy" applications into a client/server environment. Start with those that are less critical and advance slowly toward core applications. This is a good way for your IS people to cut their teeth.

In addition, you can open old data bases by connecting PCs and PC/local area networks to the mainframe. Software tools on the market will enable you to mine data and use it in ways that benefit the bottom line. More on that in upcoming columns.

So if your CIO is pressuring you to convert to the client/server religion, resist the temptation, at least until you've evaluated all the alternatives.

Put business needs first and IS needs a distant second. Failure to do so could lead to an expensive baptism. Not to mention a horde of angry shareholders at the pearly gates.

Randall K. Fields is chairman and chief executive of Park City Group, a Utah-based developer of the PaperLess Management concept, which utilizes flexible electronic management systems.