Wave Scheduler: Scheduling for Faster Turnaround Time in Peer-Based Desktop Grid Systems

The recent success of Internet-based computing projects, coupled with rapid developments in peer-to-peer systems, has stimulated interest in the notion of harvesting idle cycles under a peer-to-peer model. The problem we address in this paper is the development of scheduling strategies to achieve faster turnaround time in an open peer-based desktop grid system. The challenges for this problem are two-fold: How does the scheduler quickly discover idle cycles in the absence of global information about host availability? And how can faster turnaround time be achieved within the opportunistic scheduling environment offered by volunteer hosts? We propose a novel peer-based scheduling method, Wave Scheduler, which allow peers to self organize into a timezone-aware overlay network using a structured overlay network. The Wave Scheduler then exploits large blocks of idle night-time cycles by migrating jobs to hosts located in night-time zones around the globe, which are discovered by scalable resource discovery methods. Simulation results show that the slowdown factors of all migration schemes are consistently lower than the slowdown factors of the non-migration schemes. Compared to traditional migration strategies we tested, the Wave Scheduler performs best. However under heavy load conditions, there is contention for those night-time hosts. Therefore, we propose an adaptive migration strategy for Wave Scheduler to further improve performance.