Beyond Microservices: Testing Web-Scale RCA Methods on GPU-Driven LLM Workloads

TL;DR

This paper evaluates existing root cause analysis methods on GPU-driven large language model workloads, revealing their limitations and proposing tailored analysis techniques for improved failure diagnosis in complex LLM deployments.

Contribution

The study systematically assesses 24 RCA methods on LLM workloads, highlighting the need for specialized techniques and providing guidelines for better observability and fault localization.

Findings

Multi-source RCA approaches are most accurate.

Metric-based methods vary with fault type.

Trace-based methods largely fail on LLM workloads.

Abstract

Large language model (LLM) services have become an integral part of search, assistance, and decision-making applications. However, unlike traditional web or microservices, the hardware and software stack enabling LLM inference deployment is of higher complexity and far less field-tested, making it more susceptible to failures that are difficult to resolve. Keeping outage costs and quality of service degradations in check depends on shortening mean time to repair, which in practice is gated by how quickly the fault is identified, located, and diagnosed. Automated root cause analysis (RCA) accelerates failure localization by identifying the system component that failed and tracing how the failure propagated. Numerous RCA methods have been developed for traditional services, using request path tracing, resource metric and log data analysis. Yet, existing RCA methods have not been designed for LLM deployments that present distinct runtime characteristics. In this study, we evaluate the effectiveness of RCA methods on a best-practice LLM inference deployment under controlled failure injections. Across 24 methods (20 metric-based, two trace-based, and two multi-source), we find that multi-source approaches achieve the highest accuracy, metric-based methods show fault-type-dependent performance, and trace-based methods largely fail. These results reveal that existing RCA tools do not generalize to LLM systems, motivating tailored analysis techniques and enhanced observability, for which we formulate guidelines.