It is hard to imagine life without computers. Computers are everywhere—from small microprocessors in watches, microwave ovens, cars, calculators, and PCs, to mainframes and highly specialized supercomputers. A series of hardware and software developments, such as the development of the microchip, made this revolution possible. Moreover, computers today are rarely stand-alone devices. They are connected into networks that span the globe to provide us with a wealth of information. Thus, computers and communications have become increasingly interdependent. The nature and structure of computer networks have changed in conjunction with hardware and software technology. Computers and networks have evolved from the highly centralized mainframe systems of the 1950s and 1960s to the distributed systems of the 1990s and into the new millennium.

Today’s enterprise networks include a variety of computing devices, such as terminals, PCs, workstations, minicomputers, mainframes, and supercomputers. These devices can be connected via a number of networking technologies: data is transmitted over local area networks (LANs) within a small geographic area, such as within a building; metropolitan area networks (MANs) provide communication over an area roughly the size of a city; and wide area networks (WANs) connect computers throughout the world.

Mainframes

The parent of all computers is the mainframe. The first mainframe computers were developed in the 1940s, but they were largely confined to research and development uses. These machines were huge—in size and in price. Together with connected input/output (I/O) devices, they occupied entire rooms. The systems were also highly specialized; they were designed for specific tasks, such as military applications, and required specialized environments. Not surprisingly, few organizations could afford to acquire and maintain these costly devices. Any computer is essentially a device to accept data (i.e., input), process it, and return the results (i.e., output). The early mainframe computers in the 1950s were primarily large systems placed in a central area (the computer center), where users physically brought programs and data on punched cards or paper tapes. Devices, such as card or paper-tape readers, read jobs into the computer’s memory; the central processing unit (CPU) would then process the jobs sequentially to completion. The user and computer did not interact during processing.

The systems of the 1950s were stand-alone devices—they were not connected to other mainframes. The processor communicated only with peripheral I/O devices such as card readers and printers, over short distances, and at relatively low speeds. In those days, one large computer usually performed the entire company’s processing. Because of the long execution times associated with I/O-bound jobs, turnaround times were typically quite long. People often had to wait 24 hours or more for the printed results of their calculations. For example, by the time inventory data had been decremented to indicate that a refrigerator had been sold, a day or two might have passed with additional sales to further reduce inventory. In such a world, the concept of transaction processing, in which transactions are executed immediately after they are received by the system, was unheard of. Instead, these early computing systems processed a collection, or batch, of transactions that took place over an extended time period. This gave rise to the term batch processing. In batch jobs, a substantial number of records must be processed, with no particular time criticality. Several processing tasks of today still fit the batch-processing model perfectly (such as payroll and accounts payable).

Although the mainframe industry has lost market share to vendors of smaller systems, the large and expensive mainframe system, as a single component in the corporate computing structure, is still with us today and is not likely to disappear in the near future. IBM is still the leading vendor of mainframes, with its System/390 computers, and SNA is still the predominant mainframe-oriented networking architecture. Although IBM has been developing no centralized networking alternatives, the model for mainframe communications remains centralized, which is perfectly adequate for several business applications in which users need to access a few shared applications.

In an airline reservation database application, for example, a users’ primary goal is not to communicate with each other, but to get up-to-date flight information. It makes sense to maintain this application in a location that is centrally controlled and can be accessed by everybody. Moreover, this application requires a large disk storage capacity and fast processing— features a mainframe provides. Banks and retail businesses also use mainframes and centralized networking approaches for tasks such as inventory control and maintaining customer information.