Microservice root cause positioning method and system based on fault snapshot and large language model

By combining fault snapshots with a large language model, the problem of rapid and accurate fault location in microservice architecture is solved, achieving efficient fault repair and system stability improvement, and providing a global diagnostic foundation and deep causal reasoning capabilities.

CN122152565APending Publication Date: 2026-06-05NARI TECH CO LTD +1
View PDF 0 Cites 0 Cited by

Patent Information

Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
NARI TECH CO LTD
Filing Date
2026-01-20
Publication Date
2026-06-05

AI Technical Summary

Technical Problem

In complex microservice architectures, existing technologies struggle to quickly and accurately pinpoint the root cause of failures, especially cross-domain cascading failures. This results in long repair times, complex operations and maintenance, and reliance on expert experience. Traditional methods lack causal reasoning capabilities and data fusion mechanisms, making it impossible to effectively handle complex implicit causal relationships.

Method used

A microservice root cause localization method based on fault snapshots and large language models is adopted. The method captures abnormal alarms through a monitoring and alarm agent, aggregates multi-source telemetry data, executes an aggressive preprocessing strategy to generate structured fault snapshots, and uses a large language model trained with a specific composite to perform deep causal reasoning to generate a structured root cause analysis report.

Benefits of technology

It achieves rapid and accurate fault location, reduces fault repair time, improves system stability and reliability, with an accuracy rate of over 95%, reduces operation and maintenance costs, and can automatically handle root cause location of complex call chains.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN122152565A_ABST
    Figure CN122152565A_ABST
Patent Text Reader

Abstract

The application discloses a micro-service root cause positioning method and system based on fault snapshots and large language models, and the method comprises the following steps: an intelligent monitoring and alarming agent captures system abnormal alarms in real time; an intelligent preprocessing agent performs an aggressive preprocessing strategy on the aggregated telemetry data; the generated "fault snapshot" is taken as input and submitted to a domain-specific large language model trained in a specific composite manner; and a structured root cause analysis report is generated based on the inference result of the model. The application overcomes the context limitation of large models, realizes rapid and accurate automatic root cause diagnosis, and significantly reduces the average fault repair time.
Need to check novelty before this filing date? Find Prior Art