Performance guarantee for online deadline scheduling in the presence of overload

  • Authors:
  • Tak-Wah Lam;Kar-Keung To

  • Affiliations:
  • Department of Computer Science, University of Hong Kong, Hong Kong;Department of Computer Science, University of Hong Kong, Hong Kong

  • Venue:
  • SODA '01 Proceedings of the twelfth annual ACM-SIAM symposium on Discrete algorithms
  • Year:
  • 2001

Quantified Score

Hi-index 0.00

Visualization

Abstract

Earliest deadline first (EDF) is a widely-used online algorithm for scheduling jobs with deadlines in real-time systems. Yet, existing results on the performance guarantee of EDF are limited to underloaded systems [6,12,14]. This paper initiates the study of EDF for overloaded systems, attaining similar performance guarantees as in the underloaded setting. Specifically, we show that EDF with a simple form of admission control is optimal for scheduling on both uniprocessor and multiprocessors when moderately faster processors are available (our analysis actually admits a tradeoff between speed and extra processors). This is the first result attaining optimality under overload. Another contribution of this paper is an improved analysis of the competitiveness for weighted deadline scheduling.