Job Type Approach for Deciding Job Scheduling in Grid Computing Systems

Abstract

Problem statement: Meta-scheduling has become very important due to the increased number of submitted jobs for execution. Approach: We considered the job type in the scheduling decision that was not considered previously. Each job can be categorized into two types namely, data-intensive and computational-intensive in a specific ratio. Job ratio reflected the exact level of the job type in two specific numbers in the form of ratio and was computed to match the appropriate sites for the jobs in order to decrease the job turnaround time. Moreover, the number of jobs in the queue was considered in the batch decision to ensure server-load balancing. Results: The new factor that we considered namely, the job ratio can reduce the job turnaround time by submitting jobs in batches rather than submitting the jobs one by one. Conclusion: Our proposed system can be implemented in any middleware to provide job scheduling service.