Reproduce “FaaSRCA: Full Lifecycle Root Cause Analysis for Serverless Applications” on OpenResearch

Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.

Universal app for Apple silicon and Intel. Requires macOS 11 or later.

About this paper

Jin Huang, Pengfei Chen, Guangba Yu, Yilun Wang, Haiyu Huang, Zilong He

FaaSRCA introduces an unsupervised, full lifecycle root cause analysis method for serverless applications, integrating multi-modal observability data from both platform and application layers into a Global Call Graph. It achieved 91.54% HR@k and 94.62% NDCG@k, improving accuracy by over 20% compared to existing baselines and localizing root causes within approximately 8 milliseconds.

Key insights

  • Comprehensive root cause analysis for serverless applications necessitates integrating multi-modal observability data (metrics, logs, traces) across the entire function lifecycle, encompassing both platform and application layers.
  • Graph Attention Networks (GATs) are effective for modeling the dynamic and interconnected nature of serverless component interactions, outperforming traditional GCNs by adaptively weighting neighbor information.
  • Precisely localizing root causes requires accounting for the inherent heterogeneity of different system components, which can be achieved through mechanisms like z-score normalization on anomaly scores to compare nodes against their specific normal baselines.

Summary by alphaXiv.

Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.