Minimizing the Makespan and Economic Cost of Schedule for the Grid Applications
Avdhesh Gupta, Pankaj Agarwal, Shalini Gupta · 2013
Grid computing is a distributed computing taken to next evolutionary level. In this work, a static methodology has been adopted for defining the weights of the computational tasks and communicating edges. Also, we defined the execution time (makespan) as the total time between the finish time of exit task and start time of the entry task in the given Directed Acyclic Graph (DAG). The algorithm has been implemented for evaluation of time and cost of different random task graph or DAG of different graph size. Also, the algorithm has been executed in a grid of heterogeneous cluster of different sizes with four resources in each cluster. The primary work is to find the primary scheduling i.e., total execution time and total cost with little or no changes in primary scheduling. We have proposed an efficient scheduling algorithm, which optimize the makespan and economic cost of the schedule and minimize the requirements of processors. The algorithm has been implemented to schedule different random DAGs onto different grids of heterogeneous clusters of various sizes.