Showing papers on "Server published in 1982"

A comparison of two network-based file servers

Abstract: This paper compares two working network-based file servers, the Xerox Distributed File System (XDFS) implemented at the Xerox Palo Alto Research Center, and the Cambridge File Server (CFS) implemented at the Cambridge University Computer Laboratory. Both servers support concurrent random access to files using atomic transactions, both are connected to local area networks, and both have been in service long enough to enable us to draw lessons from them for future file servers.We compare the servers in terms of design goals, implementation issues, performance, and their relative successes and failures, and discuss what we would do differently next time.

Locating mobile servers on a network with markovian properties

Abstract: The median problem has been generalized to the case in which facilities (“servers”) can be moved, at a cost, on the network in response to changes in the state of the network. Such changes are brought about by changes in travel times on the links of the network due to the occurrence of probabilistic events. For the case examined here, transitions among states of the network are assumed to be Markovian. The problem is examined for an objective which is a weighted function of demand travel times and of server relocation costs. It is shown that when these latter costs are a concave function of the time to travel from the current server location to the new server location, an optimal set of server locations exists solely on the nodes of the network. The location-relocation problem can be formulated as an integer programming problem. The problem of locating a single server on a network which is a tree is shown to have a simple solution. A heuristic algorithm for the single server on a general network is also described. The paper develops simple bounds for the general multifacility, multistate problem and concludes with the description of a relaxed version of this model and with a discussion of the model's applicability.

Heuristic methods based on mean-value analysis for flexible manufacturing systems performance evaluation

Abstract: The application of queuing network theory to the performance evaluation of flexible manufacturing systems (FMS) usually requires the assumption of exponential processing times on machines. It has resulted in pessimistic, although qualitatively robust, evaluation algorithms so far. A recent computational technique, named mean-value analysis, allows to depart from this unrealistic assumption and to derive approximate algorithms for networks with general servers and FIFO service rule. The case of deterministic service times is particularly relevant for FMS's, and the corresponding heuristic technique is described and validated by means of simulation. This technique is liable to extend to the case of unreliable deterministic servers, thus enabling to assess the overall effect of short incidents on machines, on the production rate of the system.

The SUN workstation architecture

Abstract: The Sun workstation is a personal computer system that combines graphics and networking capabilities with powerful local processing. The workstation has been developed for research in VLSI design automation, text processing, distributed operating systems and programming environments. Clusters of Sun workstations are connected via a local network sharing a network-based file system. The Sun workstation is based on a Motorola 6800 processor, has a 1024 by 800 pixel bitmap display, and uses Ethernet as its local network. The hardware supports virtual memory management, a "RasterOP" mechanism for high-speed display updates, and data-link-control for the Ethernet. The entire workstation electronics consists of 260 chips mounted on three 6.75 by 12 inch PC boards compatible with the IEEE 796 Bus (Intel Multibus). In addition to implementing a workstation, the boards have been configured to serve as network nodes for file servers, printer servers, network gateways, and terminal concentrators. The report discusses the architecture and implementation of the Sun workstation, gives the background and goals of the project, contemplates future developments, and describes in detail its three main components: the processor, graphics, and Ethernet boards.

Distributed allocation with pools of servers

Abstract: Distributed systems make possible both a high degree of concurrency and robustness in the face of failure. One approach to achieving these goals is to employ pools of servers implementing major system functions. This paper describes the concept of pools of servers, and presents logically distributed, robust algorithms for one problem arising in this approach: the allocation of servers to clients. Three types of allocation problems are identified: free servers, preferred servers, and retentive servers. Allocation protocols based upon the idea of hash addressing are described and analyzed.

Adaptive Routing Using a Virtual Waiting Time Technique

Abstract: The virtual waiting time technique is introduced as a solution to the problem of a controller distributing work to servers of different speeds. The servers are considered to be part of a distributed system without feedback. The virtual waiting time technique is shown to minimize the average completion time for a job distributed by the controller. The virtual waiting time technique does not depend on any arrival distribution and is applicable to any service time distribution. The performance of the technique is examined for different arrival and service time distributions.

The Generalized M/G/k Blocking System with Heterogeneous Servers

Abstract: This paper studies the equilibrium behaviour of the generalized M/G/k blocking system with heterogeneous servers. Such a service system has k servers, each with a (possibly) different service time distribution, whose customers arrive in accordance with a Poisson process. They are served, unless all the servers are occupied. In this case they leave and they do not return later (i.e. they are ‘blocked’). Whenever there are n (n = 0, 1, 2,..., k) servers occupied, each arriving customer balks with probability 1 - fn+1(fk+1 = 0) and each server works at a rate gn. Among other things, a generalization of the Erlang B-formula is given and also it is shown that the equilibrium departure process from the system is Poisson.

Modeling a new technique for accessing shared buses

Abstract: Consider a queueing system in which customers (or jobs) arrive to one of Q separate queues to await service from one of S identical servers (Figure 1). Once a job enters a queue it does not leave that queue until it has been selected for service. Any server can serve any job from any queue. A job selected for service cannot be preempted. In this paper we consider jobs to be in a single class; for the multiple class result see [AFSH81a]. We assume once a queue has been selected, job scheduling from that queue is fair. In particular, our results hold for first come first serve as well as random selection [SPIR79] and, for that matter, any fair nonpreemptive scheduling policy within a queue. We assume that arrivals to each queue follow a Poisson process with the mean arrival rate to queue q being lq. The S identical exponential servers are each processing work at a mean rate of m.This system is general enough to be adaptable for modeling many different applications. By choosing the policy employed for queue selection by the servers, we can model multiplexers, channels, remote job entry stations, certain types of communication processors embedded in communication networks, and sets of shared buses. In this paper we will use the latter application to discuss a realistic situation. The elements (“jobs”) in the queues are messages to be sent from modules connected to the shared bus of the system. The servers are the buses; their service times are equal to the message transmission times. The queues are in the interface modules connected to and sharing the buses.

Data-server design issues

Abstract: The expected proliferation of local-area networks has created a need for network database servers. It is reasonable to view local-area network data servers as extensions of backend database systems. This paper addresses several critical data-server design issues: distribution of functionality, high availability, security, and performance. Particular consideration is given to applying experience with backend databases to the problems of data servers. Several design alternatives are proposed and evaluated in terms of their impact on reliability, security, and performance in a gross sense. The concluding section emphasizes the need for greater practical experience with local-area networks in order to more accurately weigh the tradeoffs of different data-server configurations.

Dispatching Service Units on Networks Using Perfect Real Time Location Information

Abstract: This paper focuses on transportation networks in which there are k — 1 stationaiiy servers and one travelling server, whose location varies deterministically over time.At some random time and at some random place (node) an incident occurs in the network. The response time is the shortest distance of a server from the incident. The analysis endsas soon as one of the servers is dispatched. Our focus is to exploit special propertiesoftheproblem formijlation to develop efficient procedures for the identification of the appropriate unit to dispatch. The moving server’s time varying, non-nodal locationadds complexity to the analysis. In application this problem often arises in urban emergencyservices. For example, it accurately depicts a situation in which one empty ambulanceisreturning from a hospital to its home satellite location and the remaining availableambulancesare at fixed known satellite locations; an ambulance dispatcher given perfect vehicular location information would be confronted with th...

An algebraic approach to the steady state solution of g/g/1//n type loops

Abstract: An explicit solution for a cyclic queueing loop consisting of two general servers and a finite number of customers with first-come-first-served behavior is derived in this dissertation, using the method of stages. Each one of these servers is represented by a vector in a finite dimensional vector space, in which pertinent properties are described. A product space is introduced together with the operational groundrules which combine the properties and interactions of these servers. In particular, the solution of this system is described in a matrix-geometric form, where the matrices are in this product space. The solution is made into a fully explicit closed form, and can be readily implemented. The form of the solution may hint as to how one may proceed to obtain an exact solution for a network of r(> 2) general servers.

The aggregate server method for analyzing serialization delays in computer systems

Abstract: The aggregate server method is an approximate, iterative technique for analyzing the delays programs encounter while waiting for entry into critical sections, non-reentrant subroutines, and similar software structures that cause processing to become serialized. The method employs a conventional product form queueing network comprised of servers that represent actual I/O devices and processors, plus additional aggregate servers that represent serialized processing activity. The parameters of the product form network are adjusted iteratively to account for contention among serialized and non-serialized customers at each physical device.

Stochastic control of a queue with two servers of different rates

Abstract: We consider the problem of optimal control of a queueing system consisting of a common queue feeding two servers of different rates. Arrivals to this system form a Poisson process and the service times are exponentially distributed. Whenever a server is idle a decision has to be made on whether to feed a customer from the queue to the idle server. The cost criterion which we desire to minimize is the average number of customers in the system or equivalently, the mean waiting time of the customers. It is shown that the optimal policy is of threshold type, i.e. the slower server should be fed a customer only when the queue length exceeds a certain threshold value.

On Computing the Stationary Probability Vector of a Network of Two Coxian Servers

Abstract: This paper presents an efficient direct method for the solution of a queueing network consisting of two Coxian servers. The structure of the transition rate matrix associated with this network is examined in some detail, and it is shown how advantage may be taken of this structure to derive the required stationary probability vector in a number of operations which is proportional to N, the number of customers in the network. The direct method which is employed, is based on a single step of inverse iteration. The method is shown to be stable.

Multi-Server Queues

Abstract: We will survey a variety of multiserver models in which the arrival stream is a Poisson process. In particular, we will consider the Erlang loss model in which arrivals finding all servers busy are lost. In this system, we assume a general service distribution. We will also consider finite and infinite capacity versions of this model. Another model of this type is the shared processor system in which service is shared by all customers.

Simple queueing systems

Abstract: Most service systems involve many servers, many queues, and many different types of customers, or objects which satisfy a conservation principle. These are likely to be interrelated in the sense that the operation of one component depends upon the others.

Statistical multiplexing with a variable number of servers

Abstract: A statistical multiplexing system that varies dynamically between a single-server system and an m-server system is analyzed. It is shown that the system has the same mean response time as a single-server system and that it betters the mean waiting time of an m-server system.

A Priority Queue with General Bulk Service

Abstract: This paper deals with priority queue with general bulk service rule. Priority and non-priority units are served by two identical servers. If a priority unit on arrival finds that (i) both the servers are busy, it balks and is considered lost to the system, (ii) one or both servers are free, it immediately goes to selvice. The non-priority units are served only when both the servers are free. Arrival of units are taken in batches of variable size whereas the service process follows general bulk service rule. Laplace transforms of generating functions of probabilities are obtained. The expressions for the steady-state probabilities and the mean queue length in the system are obtained.

Developments in the SL-10 Packet Switch: Resource, File, and Software Loading Management

Abstract: The Northern Telecom SL-10 Packet Switching System is steadily enhanced to provide better access services for users. Equally important to them, but less visible, are enhancements to improve availability and increase throughput. To this end, intelligent peripheral devices, such as storage controllers and line servers, have been added to the Switch and new strategies, especially for software loading, have been devised. Such additions have had an impact on the software architecture. The Resource Subsystem now has to cope with a wider variety of devices, including intelligent peripheral devices. The File Subsystem has been generalized to accommodate files on different types of volumes, e.g., disks and tape cartridges. A whole new Subsystem has evolved to handle strategies for loading processors in parallel and to load intelligent peripherals. The SL-10 Packet Switch software architecture, with its process environment structured in Subsystems, has proven well suited to these requirements.