Showing papers on "Shared resource published in 1979"

Fixed resource allocation method and apparatus for multiprocessor systems having complementarily phased cycles

Abstract: In multiprocessor data processing systems that utilize individual processors having interlaced cycles of operation controlled by complementarily phased clocks, contention for fixed resources such as memory, I/O, or data storage for example, may result unless an interlock method allocating the use of the fixed resource to one or another of the interlaced cycle processors is included. The usual test and set instruction approach for allocating fixed resources will not effectively interlock the use of fixed resources in a multiprocessor system having interlaced cycles without potential contention arising. The present invention resolves this difficulty by providing means for identifying which processor currently has been allocated control of a given fixed resource. In particular, in order to resolve the possible state of contention between processors accessing a common storage area which indicates the present allocation of the common fixed shared resource, the invention includes a test and set interlock instruction that is described to resolve the problem. Thus, potential contention between processors for access to the allocating means is resolved and the allocating means itself resolves potential contention for the fixed shared resource.

Computer Network Architectures

Abstract: This tutorial analyzes developments in computer network architectures from a top-down design viewpoint—starting with user interface requirements, then developing a structure to realize that interface.

Resource Sharing Protocols

Abstract: This report summarizes recent US and European work on conventions for handling two basic types of remote operations: terminal access to remote hosts and file transfer between hosts.

Deadlock Detection and Avoidance for Shared Logical Resources

Abstract: Logical resources are defined as shared passive entities that can be concurrently accessed by multiple processes. Concurrency restrictions depend upon the mode or manner in which a process may manipulate a resource. Models incorporating these single unit resources can be used to analyze information locking for consistency and integrity purposes. Mode compatibility is defined and used to derive dead-lock detection and avoidance methods. These methods generalize well-known deadlock results for single unit resources by permitting greater concurrency while still guaranteeing data consistency. This model is applicable to the standard shared (read-only) and exclusive (read-write) access modes as well as a useful subset of those proposed in the CODASYL DBMS report.

On resource sharing in a distributed communication environment

Abstract: A revolution is in the making! We are witnessing a growth rate in technological change which i s overwhelming. Thanks to enormous advances in data communications and in integrated chip technology, we are in the midst of a computer communication explosion which has already made significant changes in the field of data processing. The early phase of the revolution has passedwe have developed cost-effective data communication systems. Indeed in the last five years we have witnessed the rise of computer networks whose function it is to span intercontinental distances and provide communication among computers across nations and across the world. There now exists a large number of national networks which are in the process of interconnecting to each other in such a world network. These networks have hastened the next phase of the revolution, namely, the widespread acceptance and application of teleprocessing and networking by the business sector of our economy. As this second phase proceeds, we will see a stress placed on our computer networks in two areas. First, in the need for long-haul, wide-band inexpensive communications deep in the backbone of our networks; one answer to this need is the introduction of sophisticated packet satellite radio data communication systems. The other environment in which we will see stress is at the periphery of our networks where local accessis the major problem. The early

On channel sharing in discrete-time, multi-access broadcast communication

Abstract: : The major contribution is the novel techniques to solve a variety of distributed resource sharing problems arising in Packet Radio Networks (PRNETs). Some are applicable to a large class of resource sharing problems in computer communication networks. The results are (1) Problems of adaptive channel sharing algorithms: There are two major contributions falling under this category: (a) A novel distributed adaptive channel-access scheme, the Urn scheme, has been derived mathematically. (b) A novel mathematical approach to decentralized optimal resource sharing is developed. Using this approach, a very general characterization of optimal distributed access schemes for multi-hop networks is derived. (2) Problems of interfering queueing processes: Queueing processes may interfere with each other through their arrival processes (e.g., join the shortest queue routing) or through their service processes (e.g., destructive collisions in PRNETs). We develop novel analytical solutions, exact and approximate, to problems of interfering queues in PRNETs. (3) Capacity of multi-hop networks: We compute the capacity of tandems and show that in the limit, when the length of the tandem increases to infinity, the capacity converges to 4/27 of the bandwidth. A novel phenomena of singular topologies (i.e., where topology helps reduce interference) in PRNETs is explored.

Experience with processes and monitors in Mesa (Summary)

Abstract: In early 1977 we began to design the concurrent programming facilities of Pilot, a new operating system for a personal computer [5]. Pilot is a fairly large program itself (25,000 lines of Mesa code). In addition, it supports some large applications, ranging from data base management to internetwork message transmission, which are heavy users of concurrency (our experience with some of these applications is discussed in the paper). We intended the new facilities to be used at least for the following purposes:Local concurrent programming: An individual application can be implemented as a tightly coupled group of synchronized processes to express the concurrency inherent in the application.Global resource sharing: Independent applications can run together on the same machine, cooperatively sharing the resources; in particular, their processes can share the processor.Replacing interrupts: A request for software attention to a device can be handled directly by waking up an appropriate process, without going through a separate interrupt mechanism (e.g., a forced branch, etc.).

A system for resource-sharing in a distributed environment - RIDE

Abstract: RIDE (Resource-Sharing in a Distributed Environment) has been designed and implemented for the UNIxTM operating system to provide sharing of remote files, remote process invocation and interprocess communication in a distributed computing environment. The system is designed to support a uniform interface for both local and remote access in a network. The user programs can be executed in a single or multiple machine environment without any program modification.

Optical fibers for local computer networks

Abstract: Local computer networks that communicate over copper conductors have been developed to permit resource sharing and provide increased performance. Such networks typically operate at bandwidth-length (Bw-L) products up to a few MHz-km. To attain higher Bw-L products and enhanced EMI immunity in such net- works, an optical fiber network is advantageous. This paper describes several current system experiments that use fiber optics in local computer networks.

A file transfer protocol and implementation

Abstract: This paper presents a file transfer service in a heterogeneous computer network (HMINET). The basic architecture models are presented, the file transfer protocol is described and the implementation in the HMINET is shown with the functional capabilities and the command structure.A transfer of files between different computer system can be regarded as a service, which is divided into four steps:A: access to a file, which is administrated in a File Management System (FMS), residing on one computer systemB: transfer the file data by any transport mediumC: creation of a file copy in another EMSD: additional operations on the transferred file, e.g. print out or conversionThe usual case of file transfer (FT) between different computer systems is an offline transport by means of any storage medium (disks, tapes, etc.). Some disadvantages arise by this method:- offline transfer takes a long time- incompatibility of peripheral devices, storage medium and dat a representation can make file transfer impossible or lead to an additional overhead- for step A and C the user must be familiar with the control languages of every participated operating and file management system- additional operations on the transferred file must be organised under the system control of the destination hostA lot of advantages offers a File Transfer Service, integrated in a network, which links together the above mentioned computer systems (hosts):- with a reliable transport system in the network a fast online transfer of the file data can satisfy step B- step A and C are controlled by the File Transfer Service with only one standard interface to the user- additional operations on the copied file can he managed during the transfer of the data (conversion of file structure, file dat a and subsequent operations e.g. print)- peripheral device support by involved operating and file management system under File Transfer Service controlThe HMINET is a private star-shaped, heterogeneous network, which connects 22 process control computers and 3 mainframes. One basic function in that computer network is such a file transfer service. Typical applications of this service are:Copy of data files from and to mainframes and process control computersTransfer of programms and runable modulesRemote Job Entry and remote spool services, etc.

Shared Resources for Multiple Instruction Stream Pipelined Processors

Abstract: : This research has centered on the performance of functional resources that are used by a single multiple-stream pipelined processor. Such resources include arithmetic functional units and the modules that compose an interleaved memory. The functional requirements of such resources is that they perform some operation and resynchronize their results with the associated stream in the pipelined processor. In some instances, a replicated or pipelined resource can be used to achieve the required performance. However, in this research a simple non-pipelined unit with a fixed cycle time is investigated as a lower cost alternative. This resource is characterized by a cycle time, c, and a deadline, d, which if missed results in a penalty of one non-compute pass through the pipeline. The performance of this type of resource for various resource scheduling techniques has been determined through the use of Markov modeling and some model reduction methods. It is shown that very high performance can be obtained when effective use is made of the available deadlines. An extension to this model allows the consideration of resources with access times not equal to their cycle times. Various applications for this type of resource are examined including an implementation of a cost-effective control store which attains high performance through the use of interleaving.

File assignment in a central server computer network.

Abstract: : The placement of files among the nodes of a computer network can have a significant impact on the performance of the network. The problem of determining the optimal file placement is known as the file assignment problem. This work extends previous performance oriented file assignment research in two major areas. First, a method to obtain file assignments for read-only files that allows replication of file copies it presented. Then this method is extended to read-write files with both non-replication and replication of file copies. The network topology considered is the central server or star network topology. The star network is analyzed using techniques that model the system as a network of queues. The measure of file assignment optimality used is maximum central node throughput. For the read-only model, we introduce the concept of replicated files with split access as a method to improve central node utilization. The problem is formulated as a multiple objective integer linear programming problem. The problem is solved with a polynomially time bounded heuristic. The solution is within proven bounds on the total storage required for an optimal solution.

Dynamic simulation of a national resource sharing computer network

Abstract: A national computer network for research and educational institutions has frequently been proposed as a means for meeting selected computing needs in an effective manner. Although many technical problems remain, it is generally believed that the most difficult issues facing such a network revolve around economic, political, and organizational considerations. In order to investigate these issues, a model of a computer network was developed to test a variety of networking alternatives, and to evaluate the ways in which a network would impact its member institutions.This paper describes the use of the simulation model in a three day gaming exercise by sixteen institutional teams who made decisions about their likely participation in a network that "progressed" through several years of simulated time. Participants were able to interact dynamically with the decisions and actions of other participants, to deal with a variety of network issues, and to explore the relative advantages and disadvantages of various modes and levels of network participation.This project represented an unusual and different application of a simulation game in that it concentrated on the policy and behavioral aspects (i.e., people concerns) of what is usually considered to be a technical design problem. The players were real decision makers, playing themselves as they relate to their own institutions, rather than students. Consequently, it was possible to focus on learning about the implications of national networking, rather than merely demonstrate or teach known principles.

Simulating the behavioral implications of a national computer research network

Abstract: Publisher Summary A national computer network has often been considered as a means of sharing the diverse computing resources available at the many academic computer centers in the nation. Although such a network is technically feasible, there are many economic, organizational, and political ramifications that are difficult to predict. These unknowns present a strong inhibiting factor to potential network participants at different institutions. A simulation of the possible connection of a number of academic and research computer centers can shed light on many of the imponderables of computer resource sharing. A realistic simulation model must reflect particular site behavioral and financial constraints in addition to purely technical considerations. Indeed, it is quite likely that financial, behavioral, and political considerations may completely dominate technical considerations.