云计算环境下基于社区聚集的绿色消息系统  

Community Clustering of Green Messaging System in Cloud Computing

作　　者：李伟[1,2] 虎嵩林[1] 刘冬梅[2,3] 李锦涛[1] 

机构地区：[1]中国科学院计算技术研究所前瞻研究实验室,北京100190 [2]中国科学院研究生院,北京100049 [3]国网信息通信有限公司,北京100761

出　　处：《计算机学报》2012年第6期1327-1337,共11页Chinese Journal of Computers

基　　金：国家自然科学基金(61070027);国家自然科学基金重大国际合作项目(61020106002)资助;北京市自然科学基金(4092043);国网信息通信有限公司科技项目(SGIT[2011]567);广东省科技计划项目(2010B050100009)~~

摘　　要：云计算环境下,消息系统的客户端点部署方式对系统的性能和能耗有着重大的影响.不合理的部署会使消息系统内的路由节点承受较大的负载,增加系统的能耗.现有的部署方法往往只考虑负载均衡或者基于兴趣的部署,忽略了客户端点之间存在的复杂通信关联对系统造成的影响.文中提出一种基于社区聚集的部署方法,能根据客户端点之间的消息通信强度划分社区,并通过社区聚集的方法,提高系统性能,同时有效地降低路由节点CPU以及通信链路的能耗.实验证明该方法能有效提高消息系统的消息转发效率,并能降低路由节点的消息处理负载和通信链路中的消息量,减小系统能耗.这为绿色消息系统的研究提供了一个新思路.In cloud computing environment, the performance and energy consumption of messaging system are closely related with the deployment way of its client endpoints. An improper deployment of client endpoints brings heavy workloads to the routers in the messaging system, and increases the energy consumption. Current deployment methods just consider the load balancing, or deploy client endpoints based on their interest. The influence brought by the complex communication among client endpoints is ignored. In this paper, we propose a community-based deployment method, which extracts community structures from the client network according to the communication degrees. Through the community clustering, it improves the performance of the system, and reduces the energy consumption of the CPU of routers and communication links. The experiments show that our method can not only improve the efficiency of message delivery effectively, but also lower the load of the routers and message amount of communication links greatly, which decreases the energy consumption of the system. Our work provides a new way for the research of green messaging system.

关 键 词：云计算 消息系统 社区划分 社区聚集 能耗 绿色计算 

分 类 号：TP393[自动化与计算机技术—计算机应用技术]