Employing Checkpoint to Improve Job Scheduling in Large-Scale Systems.

The FCFS-based backfill algorithm is widely used in scheduling high-performance computer systems. The algorithm relies on runtime estimate of jobs which is provided by users. However, statistics show the accuracy of user-provided estimate is poor. Users are very likely to provide a much longer runtime estimate than its real execution time. In this paper, we propose an aggressive backfilling approach with checkpoint based preemption to address the inaccuracy in user-provided runtime estimate. The approach is evaluated with real workload traces. The results show that compared with the FCFS-based backfill algorithm, our scheme improves the job scheduling performance in waiting time, slowdown and mean queue length by up to 40%. Meanwhile, only 4% of the jobs need to perform checkpoints. © 2013 Springer-Verlag Berlin Heidelberg.