Getting to grips with client-server computing (July 1996)

Few Issues in computing cause more Confusion than client-server. It is literally impossible to open up a computing publication these days, and not find some mention of it. But just what is it? And how is it different from the styles of computing we known in the past? Here’s an abstract from ‘Client/server & Open Systems by Rand Dixon, John Wiley & Sons, New York, Jan. 1996. This abstract reprinted with permission.

Well, client-server is actually a rather abstract notion; it’s a model of computing. There are several models of computing: the old mainframe style, standalone PC, workgroup (file-serving), and, to that list, we've most recently added client-server. In the mainframe model, as you'll recall, there’s one, and only one, computer. PCs can be used to interface with the mainframe, but they're reduced to "dumb" (we should perhaps say "computationally challenged" instead - this is the politically-correct 90s after all) terminal in the process. The mainframe does everything.

Abstract part

In the client-server model, the workload is split, amongst two or more systems (not necessarily computers). This is where the abstract part comes in. The client-server computing model is defined by two or more systems in which one of the systems requests a "service" that is provided by another system. Since the model is largely implemented in software, the "systems" could quite possibly reside on the same physical computing device. Nevertheless, usually, there is a physical split: a desktop client requests services from a central computer (a "server") residing down the hall, or something like that. What kind of services are we talking about? The server might provide fax services to a department or even an entire organization. If it were an especially capable machine, the server might be used as a compute server: computationally-intensive tasks could be farmed out to this powerful device, leaving the client to take on less demanding jobs.

Corporate data

In the popular usage of the phrase, however, client-server usually means client-data-server. In this instance, corporate or departmental data would be located, processed and "served up" to clients on demand. There’s more to it than this, though. There’s an implicit requirement in the definition of client-server that the client add value to the data (if we're talking about data-serving) before presenting it to the end user. If it weren't for this proviso, file serving would qualify as client-server. "Added value" most often means presenting the data in a manner more easily digested by the user. Using a graphical user interface (GUI), like Microsoft’s Windows, IBM’s OS/2 Presentation Manager or, to a lesser extent (from a market share perspective), the OSF’s Motif, allows complex data relationships to be quickly and easily understood by end-users. But quite apart from the technical definition, there’s also a philosophical aspect to client-server. In the late 1980s, we found ourselves with networked PCs, armed to the teeth with office productivity tools.

Glass house

Just one thing was missing: access to the corporate data. That was often (usually) safely locked up on the mainframe. Without a painless way to get at it, it might as well have been on Mars. Under a banner of "information r t e people," client-server proponents now seek to tear down the barriers to the "glass house." Rather than just break through the walls of the glass house, there’s some that think we should do away with it altogether. They're the down-sizers. They reason, "now that we have client-server, why do we need mainframes?" It’s a dangerous view, however. Client-server and the mainframe style of computing we've known in the past aren't usually pitted squarely against each other; each has their strengths, each has their weaknesses, each has their place.

Click here to comment on this article

If your browser does not work with the MailTo button, send mail to with PDM_V_2.0_199607_11 as the subject.

© Oil IT Journal - all rights reserved.