Load and Thermal-Aware VM Scheduling on the Cloud Yousri Mhedheb 1 , Foued Jrad 1 , Jie Tao 1 , Jiaqi Zhao 2 , Joanna Kolodziej 3 , and Achim Streit 1 1 Steinbuch Center for Computing, Karlsruhe Institute of Technology, Germany {yousri.mhedheb,foued.jrad,jie.tao,achim.streit}@kit.edu 2 School of Basic Science, Changchun University of Technology, China scorpiozhao@yahoo.com.cn 3 Institute of Computer Science, Cracow University of Technology, Poland jokoldziej@pk.edu.pl Abstract. Virtualization is one of the key technologies that enable Cloud Com- puting, a novel computing paradigm aiming at provisioning on-demand computing capacities as services. With the special features of self-service and pay-as-you-use, Cloud Computing is attracting not only personal users but also small and middle enterprises. By running applications on the Cloud, users need not maintain their own servers thus to save administration cost. Cloud Computing uses a business model meaning that the operation overhead must be a major concern of the Cloud providers. Today, the payment of a data centre on energy may be larger than the overall investment on the computing, storage and network facilities. Therefore, saving energy consumption is a hot topic not only in Cloud Computing but also for other domains. This work proposes and implements a virtual machine (VM) scheduling mech- anism that targets on both load-balancing and temperature-balancing with a final goal of reducing the energy consumption in a Cloud centre. Using the strategy of VM migration it is ensured that none of the physical hosts suffers from ei- ther high temperature or over-utilization. The proposed scheduling mechanism has been evaluated on CloudSim, a well-known simulator for Cloud Computing. Initial experimental results show a significant benefit in terms of energy consumption. Keywords: Cloud Computing, Green Computing, Virtualization, VM Schedul- ing, Thermal-aware Scheduler, Load Balancing. 1 Introduction Cloud Computing [16,26] is a novel computing paradigm. It provisions computing ca- pacities, including hardware, software, applications, networks as well as storage, as ser- vices with a business model of pay-as-you-use. Its special features lie in that users can access the computing resources via Internet with a thin-client, such as a Web browser, without the interaction of administrators. Additionally, Cloud Computing shows the advantages in elasticity, system management, cost-efficiency, customized environment and on-demand resource provision [23,17]. Therefore, an increasing number of Cloud infrastructures [5,29,18] have been established after the first computing Cloud, the Amazon Elastic Compute Cloud [1]. J. Kolodziej et al. (Eds.): ICA3PP 2013, Part I, LNCS 8285, pp. 101–114, 2013. c  Springer International Publishing Switzerland 2013