基于设备日志的异构网络问题发现及预警方法

Problem discovery and warning method for the logs of heterogeneous network

本文主要研究在NFV场景下应对接入种类繁多的日志,基于预警规则的日志问题发现及预警方法。具体方法为基于全量日志正文数据,对细分网元的日志量进行统计,得到网元日志量监测数据和日志正文表达式监测数据,结合自动化处理与AI算法初步生成问题发现规则集和历史预警数据集,引入时间序列异常检测算法,提升问题预警的智能化水平,进一步对自动生成的规则通过专家经验进行确认,得到有效的规则集合。经过验证,采用以上方法在生产环节可得到有效规则,并生成预警数据,实现精准监测日志量异常、发现问题日志并产生预警。

This paper mainly focus on the problem discovery and warning method of log data based on warning rules for dealing with various types of logs accessed in the NFV network scenario.The specifi c method is as follows:fi rst,based on the full amount of log text data,the log volume of subdivided network elements is counted to obtain the network element log volume statistic data.After that,combined with automatic processing and AI algorithm to initially generate a set of discovery rule and a set of historical warning data.Meanwhile improves the intelligent level of problem warning based on a time series outlier detection algorithm.Further,obtains an eff ective set of rule by adding the stage of confi rming the automatically generated rules based on expert experience,so as to generate warning data based on effective rules in the production process,and realize accurate monitoring of abnormal logs,discover the problems and generate alerts.