MFRL-CA: Microservice Fault Root Cause Location based on Correlation Analysis

• Published in: 2021 8th International Conference on Dependable Systems and Their Applications (DSA) pp. 90 - 101
• Main Authors: Chen, YuHua, Chen, NingJiang, Xu, WenXiu, Lian, LinMing, Tu, Huan
• Format: Conference Proceeding
• Language: English
• Published: IEEE 01-08-2021
• Subjects: Correlation; Fault detection; Fault diagnosis; Fault propagation graph; Fault root cause location; Location awareness; Maintenance engineering; Microservice; Microservice architectures; Quality of service; Root cause analysis
• ISSN: 2767-6684
• DOI: 10.1109/DSA52907.2021.00018

With the expansion of the scale of microservice applications, its dynamic update and complex dependency call relations increase the probability of failure and the difficulty of diagnosis. How to ensure the stable operation of applications and QoS is crucial. At present, there are some researches on fault diagnosis of microservices, but there are still deficiencies in fault propagation and root cause localization. In this paper, a Microservice Fault Root cause Location Method Based on Correlation Analysis (MFRL-CA) is designed to reduce the time consumption of fault detection and root positioning. This method constructs the Microservice Fault Propagation Graph (MFPG) by collecting the correlation between dependent call data and historical fault data, and accurately infers and locates the fault causes through a new anomaly score measurement and random walk algorithm. Experiments show that this method can effectively detect faults and accurately locate the fault causes, and the accuracy is 12.75% higher than that of the benchmark method.