Characterization and enhancement of Static Mapping Heuristics for Heterogeneous Systems

Heterogeneous computing environments have become attractive platforms to schedule computationally intensive jobs. We consider the problem of mapping independent tasks onto machines in a heterogeneous environment where expected execution time of each task on each machine is known. In order to obtain insight into the performance of previously proposed static mapping heuristics, we propose the use of two performance metrics – efficacy (E) and utilization (U). Whereas there is no consistent rank ordering of the various mapping heuristics on the basis of total task completion time as task characteristics are changed, we find a very consistent rank ordering of the mapping schemes with respect to the new metrics. Using the insights provided by the new metrics, we establish a guideline that a good mapping heuristic should maximize the product E×U in order to minimize the makespan. Using this guideline, we propose approaches to improve the previously developed heuristics. Simulation results show that by using the proposed enhancements, improvements of up to 25% in makespan are achieved The new insights and approaches promises significant potential to develop better mapping schemes for heterogeneous systems.