FlowRCA: Enhancing Microservice Reliability with Non-invasive Root Cause Analysis
Zhikang Wu, Jingyu Wang, Qi Qi, Min-Gen Shu, Rui Chu, J. Li, Jing Jin, Danyang Chen · 2024
Microservice architectures, characterized by their loosely coupled services and complex call patterns, have become predominant in cloud applications, benefiting from elastic scalability and development agility. However, they face challenges in anomaly propagation and root cause analysis (RCA), often depending on the operational knowledge and system familiarity. FlowRCA, a non-invasive RCA framework, addresses these challenges by leveraging common monitoring metrics like CPU load, memory usage, and container latency to facilitate RCA in microservice environments. By analyzing causal relationships between metrics, FlowRCA clarifies fault propagation complexities, enabling accurate and comprehensive failure diagnosis. Experimental evidence shows FlowRCA’s superiority over existing algorithms, effectively identifying faulty microservices and root cause metrics in simulated environments.