WBFAS: Workflow based Failure-Aware Scheduling in Grid Computing

Abstract

Scheduling is a difficult problem in general because it is an NP-complete problem; this is true whether it is being done in Grid or in any other environment. When tasks are dependent on one another the problem becomes more complex. NP-complete problem does not have a predetermined heuristic to describe them. It's possible that a particular heuristic will function well in some circumstances but not in others, and this makes the scheduling more crucial and critical. With the goal that the application performance will be improved and the resulting throughput will be optimized, a workflow based failure aware scheduling approach (WBFAS) is proposed in this research to solve scheduling problem for dependent task in large scale system like grid computing. The workflow of dependent task is represented by directed acyclic graph (DAG). The WBFAS method is based on incremental checkpoint fault tolerant mechanism and failure information of resources. The result analysis shows that proposed method WBFAS reduces the makespan and number of failures of the system while increasing the reliability and system performance.