Linger Longer: Fine-Grain Cycle Stealing for Networks of Workstations

摘要

Studies have shown that a significant fraction of the time, workstations are idle. This paper presents a new scheduling policy called Linger-Longer that exploits the fine-grained availability of workstations to run sequential and parallel jobs. The authors present a two-level workload characterization study and use it to simulate a cluster of work stations running the new policy. They compare two variations of this policy to two previous policies: Immediate-Eviction and Pause-and-Migrate. This study shows that the Linger-Longer policy can improve the throughput of foreign jobs on cluster by 60% with only a 0.5% slow- down of foreground jobs. For parallel computing, we showed that the Linger-Longer policy outpeforms reconfiguration strategies when the processor utilization by the local process is 20% or less in both synthetic bulk synchronous and real data-parallel applications.