Scheduling Cloudlets in a Cloud Computing Environment: A Priority-based Cloudlet Scheduling Algorithm (PBCSA)

Abstract

Cloud computing is a service model that has evolved in its stature beyond its traditional bounds of infrastructure, platform and software as a service. As the surge in resource demand may hit the cloud service provider at any time, a ceaseless monitoring system is vital. The allocation of an appropriate virtual machine for the cloudlet i.e., the user workload and maintaining the work load equilibrium among the resources is the most challenging operation in the cloud environment. The proper utilization of the cloud resources can be ensured by selecting the right cloudlet scheduling and load balancing algorithm(s). The cloudlet scheduling algorithm selection is based on the combination of two or more Quality of Service (QoS) and performance metrics like makespan, throughput, cost, power consumption, virtual machine or resource utilization and load balancing etc. The load balancer module takes the responsibility of dispersing the cloudlets evenly among the virtual machines by considering various features like CPU utilization, number of processing elements, bandwidth, memory and the load limit of the virtual machines. In this paper, an effort has been made to comprehend the most persisting cloudlet scheduling and load balancing algorithms that have been proposed by the researchers. Compiling the load balancing technologies that are integrated with the contemporary cloud platforms such as Amazon Web Services (AWS), Microsoft Azure and Google Cloud Platform (GCP) has also been prioritized. This study suggests a Priority Based Cloudlet Scheduling Algorithm (PBCSA) that schedules the cloudlet according to the user priority. The Min-Min scheduler is used to schedule the high priority cloudlets and the Max-Min scheduler is used to schedule the low priority cloudlets. The experimental findings reveals that, in the majority of scenarios, the proposed algorithm outperforms the Min-Min and Max-Min scheduling in terms of makespan and virtual machine utilization ratio.