基于模糊数据挖掘的虚拟环境主机故障预测

Fault Prediction of Virtual Environment Host Based on Fuzzy Data Mining

为避免虚拟计算环境中由于资源配置不合理,导致虚拟主机服务中断或数据丢失的问题,设计一种虚拟主机故障预测模型。利用主机运行日志进行模糊数据挖掘,获得故障预测的模糊关联规则。根据故障预测中聚类区域边缘数据,给出基于规则的阈值迭代算法求解日志数据预处理修正系数,进而提高规则的匹配率。实验结果表明,该模型能够在实际服务失效前预测故障,预测准确率达到85%以上。

In order to report the service failure of the host or data interrupt in the virtual environment caused by undue resource allocation, this paper proposes a fault prediction model for virtual host. This model uses logs of the virtual host to mining the fuzzy association rules of fault predictions. Aiming at large error about rule matching problem in fault predictions caused by the edge data in cluster region, the model presents the threshold iterative algorithm based on the rules for solving the log data preprocessing coefficient,improving the rule matching rate. Experimental result shows that the prediction model can predict fault before the actual service fails,with an accuracy above 85% .