2-Phase Optimization Method for Energy Aware Scheduling of Virtual Machines in Cloud Data Centers Mohammad Moein Taheri and Kamran Zamanifar Software Engineering Department University of Isfahan, Isfahan, Iran {M.M.Taheri, Zamanifar}@ui.ac.ir Abstract—Need for computational power grows faster and faster so that we have “Cloud Computing” concept emerged from this need. In the other hand with growing popularity of computing and communication, request for more energy power increases and area of Green Computing try to moderate this procedure with revising old computing method or inventing new method to have more efficient computing material that would work more while consuming less energy and making less pollution. In this paper, the researchers tried to reduce energy consumed in Cloud computing datacenters by revising virtual machines scheduling method while keeping quality of service parameters as high as possible. We implemented our approach using CloudSim toolkit and evaluated it in compare with recent popular methods. Evaluation result demonstrates our success in reaching our aims to reduce energy consumption while keeping quality of service in acceptable range by reduction in number of virtual machines migrations. Keywords-component; Cloud Computing; Infra Structure as a Service; Virtual Machine Scheduling; Energy Efficiency; DataCenter; Green Computing; I. INTRODUCTION In recent years, emerging Cloud Computing concept besides increasing need for computing application has made a great impact on communication and computing systems. Cloud Computing provides an environment for users to access a pool of resources such as Applications, Development Environments, Virtual Machines and so on, by using network communication media to access computing resource. The superiority of this model of access lies in the way it is on demanded fashion and “Pay-per-use” model. Due to its resource sharing nature, this model can cause huge savings in resource consumption cost, maintenance cost, resource upgrading cost, etc. Moreover, it is predicted that Cloud Computing will be dominant in whole IT industry, from user application in form of web services or online storages to whole users machine as Virtual Machines that store on remote computer and user can access to that through a terminal whenever and wherever they wish to [7][8]. On the other hand, increasing demand on computing causes massive energy consumption and consequently causes pollution to the environment. R. Brown et al. in a report to USA congress in 2006 estimated energy consumption in US IT industry about 4.5 billion USD and predictions show this value will double in 2011, if it is to be continued [9]. In a research conducted in [14] on more than 5000 production servers over more than six months, they observed that in most of the times servers worked at 11-50% of machines total utilization capacity and it caused huge loss of investigation money, resource utilization, electricity power and consequently increase in total cost. Therefore, this research aims to solve this problem and reduce costs by exploiting this phenomena in Cloud Computing “Infrastructure as a Service” (Iaas) environment where users are provided with their virtual machines on their demands [8] and configuration of the virtual machines determined by an agreement on specifics of virtual machines such as amount of RAM memory, Storage capacity, amount of Bandwidth and CPU power, this agreement made between user and cloud administrator and known as “Service Level Agreement (SLA)”. So the main idea is to consolidate virtual machines in cloud data center such that virtual machines with under-utilized resources use resources in share mode and turn off extra machines to reduce energy consumption of host machines and save resources. Structure of a typical IaaS environment is presented in Fig. I-1. Other parts of this paper is organized as following: in Section 2 we will review related works in the area of energy efficient virtual machine scheduling, and then in Section 3 we introduce our proposed architecture and model and our solution to improve Figure I-1A typical Iaas Datacenter 6th International Conference on Internet Technology and Secured Transactions, 11-14 December 2011, Abu Dhabi, United Arab Emirates 978-1-908320-00-1/11/$26.00 ©2011 IEEE 525