Implementation of Fault Aware Scheduling for Reliable Execution on Desktop Grids

Abstract Desktop grids aim to make use of idle processing power as per needs of users that require resources to solve long running applications. Due to the heterogeneous environment, exceedingly dynamic resource availability, varied distribution of resources, the likelihood of faults in a desktop grid environment is much higher than conventional distributed systems. Thus, the fault tolerance becomes a necessary design feature in Desktop grid. In this paper a fault-aware scheduling is proposed in which scheduling decisions are made based upon resource reliability. The proposed strategy is compared with the traditional fault-tolerant strategy via Java based simulation toolkit Gridsim which shows better application performance and resource utilization.