Thesis (MSC) - Cyprus Internatıonal University. Institute of Graduate Studies and Research Information Technologies Department

Includes bibliography (sheets 82-93)

ABSTRACT
Cloud computing is arising as an alternative for conventional, actual or physical hardware computing in the region of parallel and cloud computing or processing. It comprises of an assortment of virtualized resources that can be provisioned on demand, contingent upon the clients' requirements. As the use of cloud services increases, the number and size of data it handles has grown exponentially and therefore there is a strong need for proper task scheduling techniques to improve efficiency in the utilization of available resources in order to achieve lesser makespan and provide a generally better quality of service. Over the decade, task scheduling became a significant research interest in various designs and areas beginning from single processor, to multiprocessor and ending with cloud computing. It is a model for activating ubiquitous network access to a shared pool of configurable computing resources or assets where accessible assets should be checked and scheduled utilizing an effective job or task scheduler to be allotted to users. The vast majority of the current job schedulers didn't accomplish the necessary standards. In this study, we propose the performance comparison of three scheduling algorithms, FCFS, SJF and RR using simulation result obtained from CloudSim environment. Our comparison revealed only slight differences between their performances depending on the metrics selected, with each having its own advantage depending on the intended objective.