computer network

[kəm′pyüd·ər ′net‚wərk] (computer science) A system of two or more computers that are interconnected by communication channels.

Computer Network

a group of interacting computer centers or systems that are linked by communication channels in order to provide fuller satisfaction of the users’ (subscribers’) information and computation needs. The creation of networks became possible as a result of advances in two areas: computers were developed that could be linked to form large complexes of computers, and new data-transmission equipment and systems were developed. Agencies, production associations, enterprises, and organizations may be users of such networks. For the network to function as an integral system, its constituent computers or computer centers must be compatible with respect to the software and hardware used and the information processed. The creation and operation of computer networks are new phases in the development of computer technology. Networks permit an efficient combination of individual and shared use of computers in the solution of a broad range of control, economic, scientific, and engineering problems.

The principal constituent elements of a computer network are shared-access computer centers, a data exchange system, and user terminal units equipped with individual or shared terminal devices.

The shared-access computer center is equipped with highspeed computers and other equipment that provide simultaneously to many users immediate access to the center’s computing and information facilities. As a result, the users can have comparatively inexpensive terminal devices instead of expensive computers that are difficult to operate. The terminals permit the users to take advantage of the large multiprocessor systems of the shared computer center to solve information and computation problems. The cost of data processing at a shared computer center is considerably lower than at conventional computer centers, and the loads computers can handle are substantially increased under conditions of shared use. At the same time, the expenditures involved in using the facilities of the shared computer center, including expenditures for terminal devices and the renting of communication lines, are less than the expenditures incurred by enterprises, institutions, and organizations that set up their own computer centers.

The computers at the shared computer center operate in the following modes: time-sharing, stack job processing of tasks received from many users through the communication channels, call-reply and man-machine dialogue. The central operating system supervises the combination of these modes at the shared computer center and carries out efficient control of the center’s facilities. The central operating system is based on the operating systems of the computers at the center. The hardware and software of the shared computer center permit a user, including a remote user, to communicate with any computer in the network.

A computer center that performs tasks for a single organization is not always maximally loaded. Experience shows, for example, that in the USSR computer centers have their maximum loads at the beginning and end of the year, half year, and quarter; at other times, the load is usually not more than half of the center’s capacity. A network permits redistribution of the load among several shared centers and allows computer facilities to be held in reserve for possible failure of one of the centers. Another important advantage of a network is that it permits the daily workloads of computer centers located in different time zones to be evened out through the transfer of part of the load to centers whose daily load peak has not yet been reached or has already passed.

One of the primary objectives of a computer network is to provide its users with the information needed to solve computation problems as well as various kinds of reference information. The information facilities of the network include data files and hardware and software for data description, collection, storage, and readout. The network’s information facilities should provide optimal conditions for centralized, integrated data processing, should ensure shared access to data that are of common interest to many users, and should protect the reliability and trustworthiness of the data received.

The most efficient way to organize the information facilities of a computer network is to make use of automated data banks, which are institutions that have their own stored information, hardware and software for data processing and storage, and special attending personnel. A computer network typically contains both local and distributed automatic data banks. The local automated data banks are set up at individual shared computer centers for the use of the subscribers served there. A distributed automated data bank makes possible the formation of interconnected data files at the various shared computer centers of the network, permits the organization of data exchange between the centers, and allows centralized control of the collection, storage, and readout of data in response to the requests of many users. The automated data bank can store not only data that are of common interest to many subscribers but also the data of individual users who find it economically advantageous to store data in the bank. An important requirement of automated data banks is that the users’ data must be protected against unauthorized access by other users and against intentional or accidental destruction or distortion (seeMEMORY PROTECTION).

The data exchange system includes data-transmission channels and equipment for interconnecting the shared computer centers and terminal units with the hardware of the data transmission system. When data are transmitted over short distances, switched and nonswitched (direct) telephone and telegraph lines are ordinarily used. The shared computer centers are linked by wide-band communication channels that permit the high-speed exchange of large volumes of data. To improve use of the communications system, increase the system’s reliability, and reduce costs, data exchange systems employ concentrators, which receive data from the users through the communication channels, group the data in packets, and forward the data to the shared computer centers. Special communications or switching processors control the data exchange process. The functions involved here include the following: organization of communication between the shared computer centers, the terminal units, and the other computer centers of the network; reception and transmission of data along communication channels; and preliminary storage and processing of data.

The structure of a computer network is largely determined by the purpose of the network, the classes of problems solved, and considerations of economy and reliability. A network may be centralized, decentralized, or mixed in structure. A centralized network has one main computer center, which controls all the facilities of the network, such as computers, programs, data, and communication channels. User tasks are executed in a centralized manner at the main center, and data common to many users are stored there. Examples of such networks are most of the networks of branch-oriented automated control systems in the USSR and the MARK-III network in the USA; these networks are constructed on the radial principle.

In a decentralized network, the shared computer centers handle the problems of their users on an independent basis, but each center has equal access to the facilities of the other shared computer centers. Such networks are generally constructed on the ring principle; an example is the ARPA Network (ARPANET) in the USA.

In mixed networks, each shared computer center functions independently in handling its users’ problems; when necessary, however, its work can be interrupted by a shared computer center of a higher hierarchic level, in order to deal with complex, higher-priority problems. The operation of such a network is coordinated by a main shared computer center, where the central supervisory service for the entire network is concentrated. Such networks are based on the radial-ring principle and thus can provide an optimal combination of individual and shared use of computer equipment.

The facilities of a computer network are efficiently managed by means of a set of programs that permits the interaction of the shared computer center in the network during the handling of interagency problems and at the same time allows the separate computer centers of the network to handle their users’ problems on an independent basis. The software of a computer network consists of system wide software and special software. The foundation of the systemwide software is the operating system of the network, which allows users to communicate with any shared computer center in the network, provides access to the operating systems of individual computers and to automated data banks and programs distributed throughout the network, ensures protection of data, and permits users to reserve their own computer equipment. The operating system of the network as a whole controls the flow of tasks entering the network, monitors the states of different elements of the network, immediately redistributes loads when overloads arise at shared computer centers and in communication channels, and organizes the parallel processing of complicated problems (with branching of algorithms) by multiprocessor systems. The special software of a computer network permits a user to solve problems by means of ready programs stored at the shared computer center or transmitted from the terminal units. In addition, the special software makes possible the compiling of new programs for user problems through the use of libraries of various types of subroutines.

Outside the USSR, computer networks have been developed in the USA (for example, ARPANET, MARK-III, and CYBERNET), Great Britain (NPL), France (Cyclada), and Japan. These networks are run on a commercial basis and are used by private and governmental organizations to solve scientific, engineering, and economic problems. In the USSR networks of computer centers are developed as the hardware for industries and branches’ automated control systems and are then integrated into the State Network of Computer Centers.

REFERENCES

Glushkov, V. M. V vedenie v ASU. Kiev, 1974.
Zhimerin, D. G., and V. A. Miasnikov. Avtomatizirovannye i avtomaticheskie sistemy upravleniia. Moscow, 1975.
Martin, J. Sistemnyi analiz peredachi dannykh. Moscow, 1975. (Translated from English.)

V. N. KVASNITSKII, V. I. MAKSIMENKO, A. L. SHCHERS, and D. G. ZHIMERIN (ed.)