Applying scheduling techniques to minimize the number of late jobs in workflow systems

  • Authors:
  • Gregório Baggio;Jacques Wainer;Clarence Ellis

  • Affiliations:
  • Universidade Estadual de, Campinas, Campinas - SP - Brazil;Universidade Estadual de, Campinas, Campinas - SP - Brazil;University of Colorado, Boulder - CO

  • Venue:
  • Proceedings of the 2004 ACM symposium on Applied computing
  • Year:
  • 2004

Quantified Score

Hi-index 0.00

Visualization

Abstract

Ordering the cases in a workflow can result in significant decrease on the number of late jobs. But merging workflow and scheduling is not trivial. This paper presents some of the problems of using scheduling results in ordering cases in a workflow and tackles two of them: the uncertainties on the cases' processing times and routing. A new approach to modeling these uncertainties is also proposed: the guess and solve technique. It consists of making a guess on the execution times and routes the case will follow, and solving the corresponding deterministic scheduling problem using a suitable technique, in this paper genetic algorithms. Simulation results show that for almost all workloads rules such as earliest due date first, and guess and solve (if the error in guessing is bound by 30%) are statistically significantly better than the commonly used FIFO rule regarding the number of late jobs.