Large Computer Networks Research Articles

An application of network management at a large computing service

Productive operation of a large computer network serving multiple user facilities in conjunction with several host sites requires the use of network management. As treated here, network management includes an array of such methods as communication systems management, problem management, change management, and inventory management. In this paper the use of some of the management techniques is described as they apply in an education support facility network. The setup of the facility is first described; then some of the techniques for managing its operation are presented.

SDS—An Innovative Student Data System

AbstractA description is presented of a comprehensive student administrative data system embedded in a large distributed processing computer network. Hardware, software, and implementation are discussed. The main subsystems include student identification, attendance accounting, class registration, and mark reporting. Various innovative aspects of the software and hardware are described. Potential further enhancements also are described.

A general class of processor interconnection strategies

A new class of general topologies is proposed in this paper for interconnecting a large network of computers in parallel and distributed environment. These structures have been shown to possess small internode distances, fairly low number of links per node, easy message routing and large number of alternate paths that can be used in case of faults in the system. The interconnection is based on a mixed radix number system, presented in this paper. The technique results in a variety of structures for a given number of processors N, depending on the required diameter in the network. A bus oriented structure is also introduced here, based on the same mathematical framework. These structures possess only two I/O ports per processor and are also shown to have small internode distances.

Toward using everybody's intelligence and knowledge

Leonard Uhr was born in Philadelphia, Pennsylvania in 1927. He received his B.A. from Princeton and Ph. D. from University of Michigan in 1957, in psychology. He was a Research Scientist at the Mental Health Research Institute and an Associate Professor of Psychology at University of Michigan, leaving in 1965 to become a Professor of Computer Sciences at the University of Wisconsin. He conducted a number of experiments on the behavioral effects of psychoactive drugs, including tranquilizers, psychic energizers and hallucinogenic agents, publishing two books: Drugs and Behavior, and Drugs and Phantasy, as well as a number of papers. He soon became interested in the potential power of the computer as a tool with which to model intellectual functions, especially perception and learning. He has written computer programs on pattern recognition, scene description, perceptual learning and language learning, using the kinds of processes that human beings seem to use. He has published over 100 papers on various aspects of this research, and also two books : Pattern and Pattern Recognition, Learning and Thought. His long-term goal is to develop a program that models the integrated wholistic cognitive functions performed by any ordinary human . being. For example, the wayan infant learns to recognize and pick up to a piece of fruit , or the way an adult learns how to handle a new set of knobs on a new stereo receiver. These appear to be simple tasks compared to playing world-championshiop chess or proving new mathematical theorems, and any non-impaired human being can handle them. But they are at the heart of human intelligence which we still know so little about. (Modeling of human cognitive functions will inevitably lead to more intelligent computers, since we can then program our understanding into the computer.) Professor Uhr is also interested in the new large arrays and networks of computers that are now being developed. These are necessary to model the brain, and to effect such highly parallel tasks as perception and recognition, or control of a motor system in real time. Finally, he has been trying to develop a self-organizing responsive computer network that would draw its users into the system, in a mutual effort to build, add to and improve upon the resources of the system. This is one example of the kind of participant-responsive system discussed in the accompanying editorial.

Optimal clustering structures for hierarchical topological design of large computer networks

AbstractLarge packet switching computer networks on the order of hundreds or thousands of nodes will soon emerge to handle the fast‐growing demands in data communication and resource sharing among various information processing systems around the world. The network topology design problem has long been recognized as extremely complex and very quickly becomes unmanageable as the size of the network increases. Existing heuristic design procedures are quite efficient for the design of small to moderate‐sized networks (25–75 nodes); however, they become very costly and even prohibitive when dealing with large networks. A design methodology based on the hierarchical clustering of the network nodes is presented in this paper in order to alleviate the computational cost involved in the design. More specifically, the emphasis is on the determination of a clustering structure which minimizes the computational cost of the design. Such a cost is assumed to have a polynomial growth with the number of nodes in the subnet to be designed. We present optimum results both for the number of clusters, number of superclusters, etc., and for the number of hierarchical levels. An expression for the average delay of a message in such a hierarchical network is also provided in terms of the average delays in the subnets composing the network. This decomposition leads to the design of smaller subnetworks for which we can utilize present design strategies.

Using encryption for authentication in large networks of computers

Use of encryption to achieve authenticated communication in computer networks is discussed. Example protocols are presented for the establishment of authenticated connections, for the management of authenticated mail, and for signature verification and document integrity guarantee. Both conventional and public-key encryption algorithms are considered as the basis for protocols.

Database management systems on a multi-computer computer

This paper discusses fitting a database and database management system onto a particular research status computer constructed of a large array or network of closely-connected component computers. The findings may have application to other distributed computer architectures.

Network usage statistics

A knowledge of the performance and reliability of large computer networks is vital for efficient operation. The kinds of statistics needed to provide this information are discussed and a description as to how they are obtained in the CSIRO network (CSIRONET) is given as a specific example. CSIRONET provides a mixed batch, interactive computing environment and hence statistics need to encompass a wide variety of peripheral devices e.g. card readers, line printers and interactive consoles. The uses to which such statistics are put aare also described. Specifically transmission line traffic and peripheral saturation may be kept under review.

Aspects of a Dynamically Adaptive Operating System

-Real-time optimization of the overall performance of a large computer system or network such as an air-traffic control system or teleprocessing network inherently requires the introduction of adaptive control into selected control functions or sets of control functions. Such adaptive control is necessitated by the high variance of the available resource inventory and the workload resource demands within the system environment. Global management of the resulting multiloop control system becomes the responsibility of the operating system. Investigated aspects of this evolutionary extension of the operating system called the dynamically adaptive operating system (DAOS) include the general methodology and the real-time modeling and estimation of resource demands. The methodology specifies the general identification, decision, and modification processes required for an adaptive operating system. Support of such processes could be attractively provided by a peripheral or miniprocessor. The problem of estimating resource service times both with and without partial service times is explored and statistical models possessing utility for real-time modeling are developed. In particular, a dynamically partitioned second moment model (DPSMM) is described. The feasibility of such models is established with respect to a General Principle of Locality. Selected simulation results are presented to evaluate selected developments with respect to an adaptive central processing unit (CPU) scheduling discipline.

Management of a large computer network simulation project

As simulation technology matures, there is an increasing demand for the development of useful products within specified time and cost constraints. Hence, simulation studies must be managed and measured by the same criteria as other large projects. The project described in this paper, a behavioral study of a possible national computer network, requires the implementation of a large and complex simulation model. In order to manage better the many individuals and diverse groups contributing to the overall effort, many of the new advances in project management, software design, and programming technology have been adopted. The results of this effort tend to support the economies and efficiencies in programming, as well as improvements in reliability, usually claimed by advocates of these techniques. More significantly, these approaches make it possible to keep all participants active in overall project progress. They have also permitted a work segmentation that takes advantage of the variety of personnel available to the project.

Plant automation in IBM: a Preface

IBM's manufacturing plants offer a valuable test ground for studying plant automation techniques. In looking for ways to improve production processes manufacturing engineers in the company's plants have experimented with many approaches to the structure and design of plantfloor computer systems. The structural solutions have ranged from dedicated stand-alone controllers scattered throughout a plant to large hierarchical networks of general-purpose computers serving an entire factory site. The implementations have included combinations of standard and modified commercially available products as well as completely “homemade” devices and software.