Integrated scheduling: the best of both worlds

This paper presents a new paradigm for parallel job scheduling called integrated scheduling or iScheduling. The iScheduler is an application-aware job scheduler as opposed to a general-purpose system scheduler. It dynamically controls resource allocation among a set of competing applications, but unlike a traditional job scheduler, it can interact directly with an application during execution to optimize resource allocation. An iScheduler may add or remove resources from a running application to improve the performance of other applications. Such fluid resource management can support both improved application and system performance. We propose a framework for building iSchedulers and evaluate the concept on several workload traces obtained both from supercomputer centers and from a set of real parallel jobs. The results indicate that iScheduling can improve both waiting time and overall turnaround time substantially for these workload classes, outperforming standard policies such as backfilling and moldable job scheduling. 1.0 Introduction Efficiently executing resource-intensive parallel scientific applications in dynamic parallel and distributed production environments is a challenging problem. Performance depends on resource scheduling at several levels: job schedulers which multiplex different applications across limited system resources and application schedulers which request resources for their applications. Applications can experience large slowdowns in time-shared systems, or high wait times in space-shared systems, due to contention for these limited valuable resources. Part of the problem is that a limited set of shared resources can rapidly become