TASK SCHEDULING IN CLOUD COMPUTING

Abstract

Recently Cloud computing has become successful and developed which, resulted many datacenter around the world. The maintenance of datacenter is necessary to the point of energy consumption and reliable operation. The datacenter maintenance can be associated with cloudlet scheduling approach by using virtualization technology. In Cloud, virtualization technology is employed to integrate physical machines into a virtual resource pool, to control resources in centralized manner. Recent work considers various strategies with only taking into account, one specific problem of task scheduling without considering the other related problems. Provisioning customer applications in the Cloud while maintaining the application’s required quality of service and achieving resource efficiency with power consumption issues are still open research challenges in Cloud computing. Hence, by considering other related problems of the task scheduling while schedule tasks can improve the resource utilization, fault tolerance, energy saving and throughput of the Cloud system. For this reason, in this thesis, we proposed a new integrated task scheduling approach that takes into account the other related issues such as VM management and datacenter management. At first level the tasks are allocated to the optimal Virtual Machine by taking into account the earliest finish time and at the second level the Virtual Machines are allocated to Server as such manner so that some machines can be switched off in order to save energy. This strategy is also helpful in improving fault tolerance capability of datacenter by switching off server when they go to high thermal state. This algorithm is implemented on CloudSim platform and the obtained experimental results show that the proposed algorithm runs efficiently, reducing the average execution time, average waiting time of tasks and improving the throughput of the Cloud system. The proposed algorithm can be easily integrated with Virtual Machine management strategies; and is fault tolerant and can efficiently deal with energy consumption problem.