Real-time anomaly detection and tracing method and system for cloud storage logs
By constructing target time-series logs and calculating the slope of proportional changes, filtering out abnormal segments, correcting timestamps, and constructing causal tracing paths, the problems of poor adaptability to load fluctuations and broken causal relationships in traditional methods are solved, and real-time anomaly detection and tracing of cloud storage logs are realized.
Patent Information
- Application Number
- CN202610273389.X
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2026-03-06
- Publication Date
- 2026-04-14
AI Technical Summary
Traditional methods for real-time anomaly detection and tracing of cloud storage logs are ill-suited to dynamic load fluctuations when faced with massive concurrent access. Fixed thresholds are difficult to adapt to, and single-frequency statistics are prone to missing abnormal behaviors. Manual retrieval leads to longer response cycles, and causal relationships are broken in a distributed environment, making it difficult to accurately depict abnormal paths and locate the source.
By constructing target time-series logs, calculating the numerical difference of failure ratios and the slope of ratio changes, filtering abnormal segments, extracting cross-node log records, correcting timestamps using the median function, and constructing a strict causal tracing path, fully automated anomaly detection and tracing are achieved.
It enables real-time anomaly detection and tracing of cloud storage logs under dynamic load, eliminates the shortcomings of fixed thresholds, breaks down information barriers in cross-node environments, demonstrates the collaborative attack transmission chain, constructs a strictly targeted tracing path, and achieves fully automated risk tracing.