🤖 AI Summary
Microservice systems are prone to cascading failures due to strong inter-service dependencies, and conventional chaos engineering relies on costly fault injection in production environments. This paper proposes a lightweight cascading failure prediction method: it automatically constructs a high-fidelity service dependency graph from distributed tracing data and performs Monte Carlo–based stochastic fault propagation simulations on this graph to enable rapid resilience assessment at the design stage—without requiring real-world fault injection. We provide the first theoretical proof that the automatically derived dependency graph supports high-accuracy resilience prediction. Evaluation on a Social Network benchmark shows prediction errors ≤ 0.0004 against ground-truth measurements; mean absolute error (MAE) is 0.025 under no-replica configurations and exactly zero with replicas, demonstrating highly accurate availability estimation capability.
📝 Abstract
Microservice applications are prone to cascading failures because of dense inter-service dependencies. Ensuring resilience usually demands fault-injection experiments in production-like setups. We propose extit{model discovery} -- an automated CI/CD step that extracts a live dependency graph from trace data -- and show that this lightweight representation is sufficient for accurate resilience prediction. Using the DeathStarBench Social Network, we build the graph, simulate failures via Monte-Carlo, and run matching chaos experiments on the real system. The graph model closely matches reality: with no replication, 16 trials yield an observed resilience of 0.186 versus a predicted 0.161; with replication, both observed and predicted values converge to 0.305 (mean absolute error leq 0.0004). These results indicate that even a simple, automatically discovered graph can estimate microservice availability with high fidelity, offering rapid design-time insight without full-scale failure testing.