Analysis and Development of Efficient Task Scheduling Strategies in Heterogeneous Cloud Environment

Abstract

In recent years, Cloud computing has become the integral part of information technology. Lots of research is being done from academic level to industry level. Cloud computing provides service to the users through internet and other distributed network environment on pay as you use basis and user demand basis. It provides an virtual environment of computing resources which can be utilized by cloud users and cloud applications. Scheduling in cloud systems is one of the biggest challenge. An efficient task scheduler is that which is flexible according to the changing environment of clouds and complexity of the submitted tasks. Efficient use of system and getting highest performance of the system is the primary goal of any task scheduling algorithm. Cloud service providers always struggles with problems such as load balancing, Task completion time and wastage of resources. This thesis basically focuses on Task completion time of tasks submitted to the virtual Machines (VMs). Multiple experiments has been performed in CloudSim 3.0.3 simulation toolkit. All the experimental results have been obtained from CloudSim by using base classes and libraries provided in toolkit. Without using any single physical machine CloudSim library gives an full environment for development and research the different techniques for simulation and modelling. Few most generic task scheduling strategies have been studied for this thesis. Based on the study a new strategy has been proposed. This new strategy is named as SCHFMC algorithm, it’s description and study has been provided in chapter 4. SCHFMC algorithm helps in allocating the tasks to the virtual machines (VMs) with varying processing capacity. It has an efficient way to utilise the full processing power of machine so that system can be active and alive without any failure. This algorithm reduces the total completion time of all tasks submitted to the virtual machines. This algorithm has performed better than generic task scheduling meth.