Automatic tuning of a fuzzy batch job scheduler using a genetic algorithm

Abstract

The paper presents the application of a genetic algorithm to automatically tune a fuzzy batch job scheduler for maximum throughput. This genetic algorithm varies fuzzy membership functions, fuzzy rules and resource limits on processors to optimize for maximum job throughput and load balancing across processors of a distributed system. Unlike most research done in the realm of load balancing and job scheduling, the paper presents an algorithm that has been evaluated in a production processing environment rather than in simulation only.