An Energy-Saving Load Balancing Method in Cloud Data Centers
With the development of virtualization technology, data center virtualization in Cloud Computing gradually become a hot topic. In the premise of ensuring users’ SLA, this paper considers the utilization of server resources, whose objective is to minimize
- PDF / 327,381 Bytes
- 9 Pages / 439.37 x 666.142 pts Page_size
- 49 Downloads / 199 Views
An Energy-Saving Load Balancing Method in Cloud Data Centers Xiao Li and Mingchun Zheng
Abstract With the development of virtualization technology, data center virtualization in Cloud Computing gradually become a hot topic. In the premise of ensuring users’ SLA, this paper considers the utilization of server resources, whose objective is to minimize the number of opening servers. We propose an energysaving strategy based on live virtual machines migration. Our ARMA-based load forecasting reduces the occurrence of virtual machines’ migration caused by instantaneous load peaks. Then we select migration virtual machines and destination servers based on our proposed algorithms. Finally, the data center reaches a load balancing state. The experiments show that the strategy can improve server resource utilization and reduce energy consumption. Keywords Cloud computing migration Energy-saving
Virtual resource scheduling
Virtual machine
35.1 Introduction Cloud computing is a hot technology [1]. In recent years, a large number of major companies’ Cloud Computing business applications are emerging. i.e. Amazon’s EC2 [2], and Microsoft’s Azure [3]. Using virtualization technology, cloud computing maps physical machines’ resources to a virtual machine layer, in which
X. Li (&) M. Zheng School of Management Science and Engineering, Shandong Normal University, 250014 Ji’nan, China e-mail: [email protected] M. Zheng e-mail: [email protected]
S. Li et al. (eds.), Frontier and Future Development of Information Technology in Medicine and Education, Lecture Notes in Electrical Engineering 269, DOI: 10.1007/978-94-007-7618-0_35, Springer Science+Business Media Dordrecht 2014
365
366
X. Li and M. Zheng
virtual machine perform users’ tasks. With the growing number of data center servers, resource scheduling problem is becoming a hotspot [4]. Cloud data center resource scheduling is divided into static scheduling and dynamic scheduling. In static resource scheduling, we decide each virtual machine running on which server. The advantage of this strategy is that the implementation is simple and reliable. However, the static scheduling scheme is not flexible. Dynamic resource scheduling is achieved by using live virtual migration. When a server is overloaded, one or some virtual machines (VM) are migrated to another server to ensure users’ SLA; When a server resource utilization is very low, we migrate all the virtual machine on it and shut down the server to achieve the purpose of saving resources. It is flexible. Dynamic resource scheduling includes four strategies, that is, load measurement strategy, information strategy, trigger strategy and migration operations strategy. Load measurement strategy is to determine the load measurement indicators. In this paper, we select CPU and memory as load measurement indicators. Information strategy determine when and how to collect the resources utilization information. In general, we collect information in the physical server and virtual machine periodically. Trigger strateg
Data Loading...
