Distributed Score Based Job Scheduling Algorithm for Cloud Computing Environment
No Thumbnail Available
Date
2016-06
Authors
Journal Title
Journal ISSN
Volume Title
Publisher
Addis Ababa University
Abstract
Cloud computing, the long-held dream of computing as a utility, is transforming a large part of the IT industry, making software even more attractive as a service and shaping the way computer hardware is designed and purchased. It is observed that the capabilities of cloud computing environment have not been used well since their core sub system, job scheduler, lacks various salient features. Consequently, this research work identified resource utilization, response time and load balancing as the major problems of job schedulers in the cloud. Thereby, we formulated a new scheduling algorithm called distributed score based job scheduling algorithm for cloud computing environment.
This algorithm mainly uses the score of resources. Processor speed, processing elements/cores, size of primary memory, size of secondary memory, and network bandwidth are parameters considered in this component. Furthermore, a component called group state manager is used to categorize resources with the intention of improving the degree of resource utilization and load balancing among resources in the cloud. In addition, it is responsible to group jobs based on their resource demand. Consolidation of jobs per resource as well as per host is also performed by a component called job consolidator. The output of this components is fed to another component called group level adaptive job scheduler. The purpose of incorporating group level adaptive job scheduler in the main architecture is to enable contextual modification of the proposed scheduling algorithm on each group. Furthermore, components dedicated for interrupt threshold calculation and job prioritization are incorporated to solve issues related to starvation.
Moreover, taking various scenarios which depict the minimum and maximum capacity of the computing environment as well as different requirements of incoming jobs we evaluated our scheduling algorithm with respect to two standard scheduling algorithms, namely First Come First Served and Round Robin job scheduling algorithms. It is observed that the proposed scheduling algorithm outperforms the counterparts from the perspective of response time, resource utilization, and load balancing.
Keywords: Cloud Computing, Job Scheduling, Server Score, Adaptive Job Scheduling, IaaS, Group Based Scheduling
Description
Keywords
Cloud Computing; Job Scheduling; Server Score; Adaptive Job Scheduling; Iaas, Group Based Scheduling