云模式下计算系统的性能监控与评价方法  

Performance Monitoring and Evaluation Methods of Computing System under Cloud Mode

作　　者：游静[1] 董小龙[1] 罗巨波[1] 孙玉强[1] 

机构地区：[1]常州大学信息科学与工程学院,常州213164

出　　处：《系统仿真学报》2013年第12期2845-2850,共6页Journal of System Simulation

基　　金：江苏省自然科学基金(BK2009535);江苏省高校自然科学基金(07KJB520022);常州市青年人才基金(CQ20100007)

摘　　要：用户可以基于云计算平台构建企业应用,并对外提供服务。这种应用系统称之为云模式下的计算系统。用户希望该系统能够和本地企业服务器一样可监可控甚至更方便。为此,通过搭建仿真环境,模拟了基于云计算平台提供的企业服务;设计了云模式下计算系统的性能监控工具,对常用的Linux和Windows虚拟计算系统资源进行自主监测;提出了云模式下计算系统性能的动态评价方法,将用户最直接的体验——请求的响应时间——作为性能数据,建立了响应时间和多种计算资源之间的直接关系,并采用移动窗口方式对系统性能进行持续监测和评价;针对实验中采集的资源和性能数据,获得了响应时间对多种资源的多阶段基因表达式编程模型,结果表明,理论响应时间较好地反映了系统的运行状态和性能变化趋势。Users can build the enterprise applications based on the cloud computing platform, and provide services for their users. This kind of application system is called computing system under cloud mode. Users wish that the system can be monitored and controlled the same as local servers and will be even more convenient. Therefore, a simulation environment was built to simulate the enterprise services based on cloud computing platform in this paper. A performance monitor tool under cloud mode was developed to monitor independently the popular virtual computing system resource such as Linux and Windows. And then a dynamic evaluation method of computing system under cloud mode was put forward, in which the response time of requests that was the most direct user experience was taken as performance data. This method established a direct relationship between the response time and a variety of computing resources, and it could monitor and evaluate continuously the system performance by moving window. Finally, for the data collected in the experiment, a multi-stage gene expression programming model of the response time to resources was obtained. And the result shows that the theoretical response time well reflects the running state and performance trends.

关 键 词：性能监控 性能评价 云模式 基因表达式编程 系统可靠性 

分 类 号：TP302.7[自动化与计算机技术—计算机系统结构]