Network security log analysis and incident response based on Drain3 and Loganomaly
As network security issues become increasingly critical with the development of information tech-nology,this study focuses on analyzing the logs of HDFS-v1 to ensure the stable operation of network sys-tems.Firstly,the log is parsed by using Drain3 algorithm,which effectively avoids constructing a very deep and unbalanced tree.Experiment results show that its precision,recall,F1 score and accuracy are as high as 100％.Then,anomaly detection is carried out based on Loganomaly algorithm.The loss value of train-ing set and validation set of training results is 0.21 and 0.18,respectively.The accuracy of prediction re-sults is 96.889％,the recall rate is 93.604％,and the F1 metric is 95.218％.After that,the system uses the Drain3 algorithm to classify anomaly logs.Finally,the system uses remote control to implement rapid response to abnormal events and sends alarm emails to ensure rapid and effective fault handling in case of HDFS emergencies,ensuring the stable progress of big data processing tasks.
NETL
NSTL
万方数据
