Fuzzing Microservices in Face of Intrinsic Uncertainties

2026-03-02

AI Summary

This work proposes a novel uncertainty-driven system-level fuzz testing paradigm for microservice systems, which inherently exhibit multidimensional uncertainties due to dynamic scaling and decentralized control. These uncertainties render conventional single-service testing approaches ineffective in capturing non-deterministic behaviors, performance fluctuations, and fault propagation. By modeling and injecting uncertainty as the core driving force, the proposed method integrates causal inference to trace uncertainty propagation paths, enabling precise fault localization and quality assessment. The framework leverages service virtualization, uncertainty simulation, and adaptive test generation to construct a system-level testing architecture compatible with continuous integration pipelines. Empirical validation on an e-commerce microservice system demonstrates its effectiveness, offering a scalable and automated foundation for enhancing the reliability and resilience of industrial microservice deployments.

Technology Category

• Multiagent Systems: Multiagent Systems under Uncertainty
• Reasoning under Uncertainty: Uncertainty Representations
• Machine Learning: Calibration & Uncertainty Quantification

Application Category

• Systems and Infrastructure for Web, Mobile and WoT: Web performance, measurement, and characterization
• Security and Privacy: Large-scale security measurements
• User Modeling, Personalization and Recommendation: Federated recommendation systems and personalization

Abstract

The widespread adoption of microservices has fundamentally transformed how modern software systems are designed, deployed, operated and maintained. However, well-known microservice properties (e.g., dynamic scalability and decentralized control) introduce inherent and multi-dimensional uncertainties. These uncertainties span across inter-service interactions, runtime environments, and internal service logic, which manifest as nondeterministic behaviors, performance fluctuations, and unpredictable fault propagation. Existing approaches do not have sufficient support in capturing such uncertainties and their propagation in industrial microservice systems, and these approaches mostly focus on single-service testing. In this paper, we argue for a novel paradigm: ``uncertainty-driven'' and ``system-level'' microservice testing. We outline key research challenges, including the modeling and injection of uncertainties and their propagation, causal inference for fault localization, and multi-dimensional analyses and assessment of uncertainties and their impact on system quality. We propose an architecture for continuous uncertainty-driven and system-level microservice fuzzing, which integrates service virtualization, uncertainty simulation, adaptive test generation and optimization\revision{, and illustrate it with an e-commerce example we developed}. Our goal is to inspire the development of scalable and automated system-level testing methods that improve the dependability and resilience of industrial microservice systems, with the explicit consideration of uncertainties and their propagation.

Problem

• microservices
• uncertainties
• system-level testing
• fault propagation
• nondeterministic behaviors

Innovation

• uncertainty-driven testing
• system-level fuzzing
• microservice resilience
• adaptive test generation
• fault propagation analysis