Job Scheduling Strategies in Grid Computing

Abstract

Grid computing can be thought of as large-scale distributed cluster computing and distributed parallel network processing. Users can obtain enormous computing power through network technology, which is challenging to get from a single computer. Job scheduling in grid computing is a critical issue that affects the overall grid system capability. In traditional scheduling, jobs are placed in queues, waiting for the availability of resources. Reservations reject if the required resources not obtained at the specified time. The impact that arises is the reduced use of resources. The scheduling algorithm and the parameters used to perform the work may vary, such as execution time, delivery time, and the number of resources. There is no guarantee when the job will execute using the scheduling algorithm. Therefore, it is necessary to improve resource utilization in the grid system and ensure that jobs will be carried out. This paper proposes a reservation scheduling strategy for MPI work, First Come First Serve Left Right Hole (FCFS-LRH). MPI jobs execute simultaneously, using more than one resource for implementation. When Completed, user MPI jobs will be scheduled on virtual compute nodes and mapped to actual compute nodes. The experimental results show that the increase in resource utilization strongly influenced by time flexibility.