Today's microservice-based architectures produce significant amounts of telemetry data, and detecting anomalies and performing root cause analysis have become challenging tasks. The conventional approach to monitoring does not effectively identify the dynamic relationships between microservices and temporal patterns of system behavior. This paper presents a spatiotemporal anomaly detection technique that utilizes a combination of Graph Neural Networks, Long Short-Term Memory, and Variational Autoencoders. The proposed technique effectively detects anomalies and identifies the reasons behind them. Additionally, this paper introduces application-level features, such as HTTP request information, to provide a better understanding of the reasons behind anomalies. The results of the experiments prove that the proposed technique effectively detects anomalies and identifies the reasons behind them. Index Terms—Anomaly Detection, GNN, LSTM, VAE, Microservices, Root Cause Analysis.
Shariff et al. (Fri,) studied this question.