Sever Performance Degradation Analysis Based on Average Load Chaotic Time Series Forecast

摘要：

A long-running Web software system may lead to the exhaustion of resources, which cause performance degradation. To solve that problem, needs to predict the crucial resources using situation, and then carry out the proper software rejuvenation strategies. At first, this paper identify the average load chaotic character which can be described by using G-P algorithm to analyze correlation dimension changing with embedding dimension, then get the largest Lyapunov exponent through small data method and build chaotic time series prediction model based on largest Lyapunov exponent for average load time series. The experimental results show that the prediction model can precisely make short-time prediction to the Web servers load, which can efficiently estimate the performance degradation situation and provide foundation for the software rejuvenation.

关键词： performance degradation time series average load software rejuvenation

作者: Junwei Ge Shanfeng Chen Yiqiu Fang

作者单位: College of software Chongqing Univ.of Posts and Telecom.Chongqing 400065, China College of Computer Chongqing Univ.of Posts and Telecom.Chongqing 400065, China

会议类型: 国际会议

会议名称: The 2010 International Conference on Computer Application and System Modeling(2010计算机应用与系统建模国际会议 ICCASM 2010)

会议地点: 太原

会议语种:英文

页码: 220-223

在线出版日期: 2010-10-22（万方平台首次上网日期，不代表论文的发表时间）