Adaptive Deadline Based Dependent Job Scheduling algorithm in cloud computing

Abstract

Cloud computing is a platform for hosting an immense number of applications and computing services that will continuously change due to a large number of jobs randomly submitted by the end user. Thus, scheduling becomes strenuous in cloud computing because of an immense number of jobs submitted randomly. The ultimate intention of the proposed work is to minimize the makespan of the job, to improve the processor utilization irrespective with the cloud environment. Hence, this paper posits a novel approach called Adaptive Deadline Based Dependent Job Scheduling (A2DJS) algorithm in cloud computing that comprises of three major components as job manager, data center and VM creation. Here, the job manager embeds with dependency resolver and task-prioritizer. The dependency resolver will determine the dependency among the tasks and task-prioritizer will prioritize the tasks to avoid starvation. Moreover, the data center embeds with job scheduler and host creation with VM allocation. The job scheduler schedules the job with the VM existing. The host creation with VM allocation allocates the jobs to the VM in a two-tier VM architecture. This contribution will mitigate the makespan of the job, evade starvation and improve the processor utilization. The results are simulated using cloudsim that shows the performance of the proposed A2DJS algorithm better when compared to the existing algorithms.