MicroRCA Root Cause Localization of Performance Issues in Microservices



核心做法是训练一个AE,学习指标的正常模式。认为对reconstrtuction error贡献越大的指标越异常


We observe that applying autoencoder on the service relevant metrics can significantly improve the accuracy of culprit service localization by ranking the faulty service within the top two. Table 2 shows the overall performance of the above two methods for all anomaly cases. It shows that complementing culprit service localization with autoencoder can achieve a precision of 92%, which outperforms 61.4% than the results of CSL only.

Last update : February 13, 2023
Created : February 13, 2023