Search papers, labs, and topics across Lattice.
This paper introduces MetaRCA, a framework for root cause analysis in cloud-native systems that addresses limitations in scalability, generalization, and domain knowledge integration of existing causality-based methods. MetaRCA constructs a Meta Causal Graph (MCG) offline by fusing knowledge from LLMs, historical fault reports, and observability data, and then performs lightweight online inference by instantiating and pruning the MCG based on real-time data. Experiments on public and production failures demonstrate that MetaRCA achieves state-of-the-art performance, outperforming baselines by significant margins in both service-level and metric-level accuracy, while also exhibiting robust cross-system generalization.
LLMs and historical data can be combined to create a reusable knowledge base that slashes the cost of root cause analysis in complex cloud systems, improving accuracy by up to 48%.
The dynamics and complexity of cloud-native systems present significant challenges for Root Cause Analysis (RCA). While causality-based RCA methods have shown significant progress in recent years, their practical adoption is fundamentally limited by three intertwined challenges: poor scalability against system complexity, brittle generalization across different system topologies, and inadequate integration of domain knowledge. These limitations create a vicious cycle, hindering the development of robust and efficient RCA solutions. This paper introduces MetaRCA, a generalizable RCA framework for cloud-native systems. MetaRCA first constructs a Meta Causal Graph (MCG) offline, a reusable knowledge base defined at the metadata level. To build the MCG, we propose an evidence-driven algorithm that systematically fuses knowledge from Large Language Models (LLMs), historical fault reports, and observability data. When a fault occurs, MetaRCA performs a lightweight online inference by dynamically instantiating the MCG into a localized graph based on the current context, and then leverages real-time data to weight and prune causal links for precise root cause localization. Evaluated on 252 public and 59 production failures, MetaRCA demonstrates state-of-the-art performance. It surpasses the strongest baseline by 29 percentage points in service-level and 48 percentage points in metric-level accuracy. This performance advantage widens as system complexity increases, with its overhead scaling near-linearly. Crucially, MetaRCA shows robust cross-system generalization, maintaining over 80% accuracy across diverse systems.
