The Impact of More Accurate Requested Runtimes on Production Job Scheduling Performance

This paper addresses the question of whether more accurate requested runtimes can significantly improve highperformance priority backfill policies, for production workloads running on leading edge systems such as the 1500 Origin 2000 system at NCSA or the new TeraGRID. This question has been studied previously for FCFS-backfill using a limited set of performance metrics. The new results for higher performance backfill policies, heavier system load, and for a broader range of performance metrics show that more accurate requested runtimes have much greater potential to improve system performance than suggested in previous results. Furthermore, the results show that (a) using user test runs to improve requested run time estimates can achieve most of the benefit, and (b) users who provide more accurate requested runtimes can expect improved performance, even if other users do not provide the more accurate requests.