Modified Min - Min Heuristic for Job Scheduling Based on QoS in Grid Environment

Job scheduling is an influential component of Grid computing. Grid consists of large number of heterogeneous resources for solving large scale problems in the area of engineering and research discipline. So a proper job scheduling mechanism is essential for the efficient and reliable working of Grid because if a job is submitted to an inappropriate resource then it will increase the completion time of the job as well as decrease the overall performance of Grid, e.g. if a communicational intensive job is submitted to low bandwidth resource, then it will result in more communication time and thereby delaying the overall execution. Major goals of scheduling is to achieve better throughput while matching applications with the available number of resources. In this paper a modified QoS guided job scheduling algorithm is proposed. We divide the resources in the four different classes High QoS in term of processing speed, Low QoS in term of processing speed, High QoS in term of bandwidth, Low QoS in term of bandwidth. After choosing appropriate class based upon the job parameters, min-min heuristic schedules the job onto the best available resource of that class. The algorithm is tested in a simulated Grid environment. The output of algorithm shows that the new modified QoS guided Min-Min heuristic give much better result as compare with the traditional one.