Distributed gang scheduling in networks of heterogenous workstations

  • Authors:

  • Khaled Al-Saqabi;Syed Sarwar;Kassem Saleh

  • Affiliations:

  • Department of Electrical and Computer Engineering, Kuwait University, P.O. Box 5969, Safat, Kuwait;Department of Electrical Engineering, University of Portland, Portland, OR 97203-5798, USA;Department of Electrical and Computer Engineering, Kuwait University, P.O. Box 5969, Safat, Kuwait

  • Venue:
  • Computer Communications
  • Year:
  • 1997

Quantified Score

Hi-index 0.24

Visualization

Abstract

The wide availability of workstation networks and the rapid evolution of workstation technology is a motivation for investigating methods of harnessing the full power of such systems. Individual workstations are not usually effectively utilized by their owners. Owners may be willing to lend the processing power of their workstations if used in an unobtrusive way. The ability to effectively borrow the idle cycles of the workstations in a network and efficiently schedule parallel application programs concurrently onto those idle workstations is the topic of this paper. In this paper, we present a distributed scheduling algorithm that will track the available workstations, i.e. workstations not used by their owners, in networks and act upon those workstations by scheduling processes of parallel applications onto them. Our scheduling objectives are minimizing the average Turn Around Time (TAT) of the scheduled applications and maintaining fairness among scheduled applications by granting each application all the resources it requires. Moreover, scheduling solutions are narrowed to those that produce a responsive and scalable scheduling algorithm.