下载此文档

基于日志信息的故障诊断分析word论文.docx


文档分类:论文 | 页数:约63页 举报非法文档有奖
1/63
下载提示
  • 1.该资料是网友上传的,本站提供全文预览,预览什么样,下载就什么样。
  • 2.下载该文档所得收入归上传者、原创者。
  • 3.下载的文档,不会出现我们的网址水印。
1/63 下载此文档
文档列表 文档介绍
Abstract
In the puter system, log is usually to be the first source for obtaining the system situation and diagnosing the failure of system. However, with the development of puting and cluster environment, the architecture of system and software is plicated. munication between different layer of hardware and software in the system is ing more frequency. It enhances the coupling of system and makes fault diagnosis using system logs difficult. The traditional system fault diagnosis based on log relies on artificial analysis. It is difficult to analyze system in real-time prehensive, when dealing with massive log generated by ponents in modern system. Therefore, it is a significant meaning for current management puter system that to design a log management and analysis system.
The unified management and analysis system for log (UiLog) collect fault logs gener- ated by ponent of cloud system and provide management and analysis functions. When a failure occurs, UiLog will determine the fault catalog and sort the entire fault logs in causality to help administrator for fault diagnosing. Firstly, UiLog manage all the logs in the system to ensure UiLog can obtain the information of fault. Secondly, UiLog take ad- vantage of new fault classification measure to reduce the dependency of human knowledge library. Through the Fault Keyword Matrix, UiLog can classify fault log efficiency. On the other hand, UiLog improves the traditional fault correlation analysis based on time. Using the result of fault classification to determine the size of time windows, it can improve ac- curacy of fault classification and help administrator for finding the root cause of fault.
The result of evaluation shows that UiLog can manage logs of entire system and clas- sify logs by fault type. Meanwhile, UiLog also can find root-cause of fault. For log classi- fication, the new approach improves the accuracy of fault classification to 95%. For log correlation, the truncation fault rate and collisio

基于日志信息的故障诊断分析word论文 来自淘豆网m.daumloan.com转载请标明出处.

相关文档 更多>>
非法内容举报中心
文档信息
  • 页数63
  • 收藏数0 收藏
  • 顶次数0
  • 上传人wz_198613
  • 文件大小937 KB
  • 时间2018-02-24