Computer scheduling methods and their countermeasures

  • Authors:

  • Edward G. Coffman, Jr;Leonard Kleinrock

  • Affiliations:

  • Princeton University, Princeton, New Jersey;University of California, Los Angeles, California

  • Venue:
  • AFIPS '68 (Spring) Proceedings of the April 30--May 2, 1968, spring joint computer conference
  • Year:
  • 1968

Quantified Score

Hi-index 0.00

Visualization

Abstract

The simultaneous demand for computer service by members from a population of users generally results in the formation of queues. These queues are controlled by some computer scheduling method which chooses the order in which various users receive attention. The goal of this priority scheduling algorithm is to provide the population of users with a high grade of service (rapid response, resource availability, etc.(, at the same time maintaining an acceptable throughput rate. The object of the present paper is to discuss most of the priority scheduling procedures that have been considered in the past few years, to discuss in a coherent way their effectiveness and weaknesses in terms of the performance measures mentioned above, to describe what the analysis of related queueing models has been able to provide in the way of design aids, and in this last respect, to point out certain unsolved problems. In addition we discuss the countermeasures which a customer might use in an attempt to defeat the scheduling algorithm by arranging his requests in such a way that he appears as a high priority user. To the extent that we can carry out such an undertaking, the single most important value of this consolidation of the results of analysis, experimentation, and experience will be in the potential reduction of the uncertainty connected with the design of a workable service discipline.