Showing papers on "Server published in 1976"

A principle for resilient sharing of distributed resources

Abstract: A technique is described which permits distributed resources to be shared (services to be offered) in a resilient manner. The essence of the technique is to a priori declare one of the server hosts primary and the others backups. Any of the servers can perform the primary duties. Thus the role of primary can migrate around the set of servers. The concept of n-host resiliency is introduced and the error detection and recovery schemes for two-host resiliency are presented. The single primary, multiple backup technique for resource sharing is shown to have minimal delay. In the general case, this is superior to multiple primary techniques.

On Partial Equilibrium in a Queuing System with Two Servers

Abstract: This paper will deal with the following question: what happens in a service system where the service is supplied by two competitive firms? By a service system we mean any system where customers have to wait in lines before they are served. As we will show, we cannot apply the usual results of economic theory to service systems. For example, in ordinary economic theory two firms supplying the same good will always sell it at the same price. This result is not always true for the case of service and waiting lines. We will discuss the conditions under which two service firms selling at the same price are not in equilibrium. A discussion of queuing systems with single server (that is, a monopoly) are found in the works of Edelson [1], Naor [3], Knudsen [2] and others. However, there is no mention in the literature of a general equilibrium in a queuing system with more than one server. Between these two extreme cases, the general equilibrium and the determinate single-firm model, lie the partial equilibrium cases, and this paper deals with one of these.

Some optimal control problems for queueing systems

Abstract: We consider here the optimal control of the number of servers in the M/M/S queue and M/G/1 queue. There are switching costs for every change of the control variable and the finite horizon problem is considered.

Experience with the central server model on a lightly-loaded system

Abstract: Analytic models potentially provide an elegant means of estimating the performance of computer systems at a rather nominal cost. The assumptions which must be made about a system in order to construct a mathematically tractable model, however, are usually severe enough to cast doubt on the validity of the results produced by such a model. Further, relaxation of an assumption often renders an analytic technique inapplicable, such that simulation becomes the preferred approach to system modeling.To investigate the boundary between analytic models and simulation models a computer program has been developed which provides a framework within which to develop and compare both analytic and simulation models. A general queuing network architecture has been adopted. The program is extensible with respect to analytic techniques, queuing disciplines and probability density functions.Using this program, models of a lightly loaded batch multiprogramming system were constructed. The system was viewed as a network of servers preceded by queues. The central processing unit was one server and each I/O channel was treated as a separate server. The channels were organized in parallel with respect to each other, and in series with respect to the central processing unit. Tasks were viewed as tokens flowing through the network, being delayed by each server for queuing delays and service. Delays between task termination and initiation of a new task were modeled by additional servers operating in parallel to the I/O channel servers. An analytic solution was computed for this model for the steady state case with exponentially distributed service times and the first-come-first-served queuing discipline with unlimited capacity queues. Typical errors of 18% were observed for cpu utilization and I/O activity. The model was then modified to more realistically account for the behavior of partitions. The delay servers were modeled with unit capacity queues. This model was solved by simulation, yielding errors of 4% for cpu utilization and I/O activity.

The allocation of non-identical machines among non-identical servers

Abstract: The problem of allocating non-identical machines among non-identical servers is considered under steady-state conditions for the case of quasirandom input and exponential service times. Machines are assigned to operators with the objective of minimizing an expected cost model of the queueing system. Different classes of machines have different service rotes and, for a given class of machines, the service rates are different among servers. A pattern search procedure is used to solve the resulting optimization problem. Computational experience is presented, together with an analysis of the sensitivity of the model to errors in estimating the values of the parameters of the model.

Models in Resource Allocation — Trees of Queues

Abstract: Queueing models play an important role in resource-allocation problems. In this paper a tree consisting of terminals (sources of requests), queues and synchronous servers is investigated. The generating function and the first moment of the holding time of a request generated by an arbitrary terminal are calculated.

A heterogeneous system with finite waiting space

Abstract: A simple queue with two heterogeneous servers is analyzed. The emphasis is on comparing the two-server heterogeneous and homogeneous systems with the restriction of an upper limitN on the queue size. The optimal service rates for both the servers are found in terms of the arrival rate and the traffic intensityρ. The average characteristics of the heterogeneous system are minimized, and their improvement over the corresponding homogeneous system characteristics is established. For different values ofN andρ, tables are given which compare the average characteristics of the two systems.

Design considerations for a user oriented discrete-event simulation language

Abstract: Discrete event simulation languages can be described in terms of simulation perspective and language structure. A useful language should allow an emphasis to be placed on transactions, queues, and item flow. The language structure should be self-documenting, readable, and descriptive rather than algorithmic in philosophy. SIMFO, a newly designed language, provides the necessary definition of a system using a COBOL-like structure, and complying to the above requirements. A system can be defined in terms of the items, facilities, servers, queues, and item flow. Branching involves only an IF - - - THEN construction. Aside from summary information, all analyses are performed off-line.

Signal processing system

Abstract: PROBLEM TO BE SOLVED: To appropriately perform request regulation of a system comprising a plurality of signal processing servers and a plurality of data processing servers.SOLUTION: A congestion determination device 50 in a system instructs a signal processing unit 200 to perform request regulation when congestion is detected in a data processing server. The congestion determination device 50 terminates the request regulation when a request number condition sufficiency notification with regard to a regulation target request number in the signal processing server of the signal processing unit 200 is received after the request regulation is started and a resource usage rate in the data processing server becomes lower than a prescribed threshold. A representative server of the signal processing unit 200 estimates a regulation target request number for the whole signal processing unit 200 on the basis of the regulation target request number received in the own server, estimates a data request number for the whole data processing server from the estimated value and transmits the request number condition sufficiency notification to the congestion determination device 50 when the estimated value is less than a prescribed threshold C.