Joint Temporal-Structural Representation Learning for Distributed Fault Discrimination in Microservice Architectures

2026-05-03

AI Summary

This work addresses the challenge of accurately identifying distributed failures in microservice systems, where diverse failure modes, complex dependencies, and time-varying states hinder reliable diagnosis. To this end, the authors propose a unified dynamic graph learning framework that models microservice execution as a sequence of temporal graphs. By integrating multi-source observation alignment, structure-enhanced temporal encoding, and attention-based message passing, the framework jointly captures both topological structures and their temporal evolution. A dual readout mechanism is further introduced to aggregate node-level and global temporal information for system-level fault classification. Notably, this approach is the first to unify dynamic topology modeling and temporal dynamics within a single framework, significantly enhancing discriminative robustness under complex noise and interaction scenarios, and outperforming existing baselines across multiple evaluation metrics.

Abstract

Addressing the diverse fault morphologies, complex dependencies, and time-varying operational states in microservice distributed systems, this paper proposes a distributed fault discrimination model based on temporal graph neural networks. This model characterizes the microservice operation process as a dynamic graph sequence evolving, and performs joint representation learning of temporal modeling and structural interactions within a unified framework. First, service-level multi-source observation signals are aligned and characterized to construct node feature sequences and their corresponding time-dependent dependencies. Then, a temporal coding module is introduced to extract the dynamic evolution representation of service states, and at each time step, attention-based structured message passing is used to characterize dependency interactions and propagation associations, forming a structure-enhanced temporal node representation. Furthermore, a dual readout mechanism is employed to aggregate the node and temporal dimensions, obtaining a system-level global representation and outputting the fault category distribution. Finally, supervised learning objectives are used to optimize model parameters, enabling the model to learn stable discrimination evidence under complex interactions and multi-source noise conditions. Comparative experimental results show that the proposed method achieves superior performance on multiple evaluation metrics, validating the effectiveness of jointly modeling temporal evolution and dependency structures in improving the distributed fault discrimination capability of microservices.

Problem

• microservice
• distributed fault discrimination
• temporal dynamics
• structural dependencies
• fault morphology

Innovation

• Temporal Graph Neural Networks
• Joint Representation Learning
• Dynamic Graph Sequence
• Attention-based Message Passing
• Distributed Fault Discrimination