Autonomic Resource Allocation for Cloud Computing

Abstract

Cloud Computing is a vast distributed Computing environment, in which so many concepts such as virtualization, scalability is explained. As computing infrastructure is expanding day-by-day due to arrival of different tasks over the internet, it results in overloading, so in order to avoid overloading, Poorly performing VMs are migrated to other VMs, which are capable of handling overloading, and under loading. We will make use of MOTS sub-system for effective task scheduling and resource allocation and, we have compared the results with different existing task scheduling techniques such as Round Robin(RR) and First Come First Serve (FCFS). We have Simulated the Results by using CloudSim tool and, further we have extended the work for effective task scheduling with the help of our proposed Multi-Objective Task Scheduling Algorithm(MTAA). In this thesis, we are proposing a optimal resource management
system (AS-ROM), which addresses two major topics of cloud environment, they are resource allocation and workload prediction. In this system we are proposing Multi-Objective Task Scheduling optimization (MOTS) which covers resource estimation and resource allocation. It determines the optimal pattern over VMs by considering scheduled tasks, task execution cost, execution time, the length of the task over VMs and power consumption. In this thesis, we described about DBN-workload prediction model which considers the probabilities of various tasks, allocated over every virtual machine and, this model depends on maximum memory utilization, and expected time to compute(ETC), with the help of this model resource discovery and resource allocation are predicted by making use of MOTS(Multi-Objective Task Scheduling)Algorithm.