Design and Implementation of an Intelligent Operation and Maintenance System Based on Distributed Tracing

This paper proposes an intelligent operation and maintenance system based on distributed tracking technology, which integrates the Jaeger distributed tracking system and Prometheus monitoring system to build an operation and maintenance solution that includes three core functions: data collection, anomaly detection, and fault alarm. The system adopts dynamic baseline algorithm combined with quantile analysis to realize service call time-consuming anomaly detection, and designs multilevel alarm strategies to improve the effectiveness of alarms. Experiments show that the system exhibits excellent performance under different fault types. The F1scores of database, network, CPU, and memory faults are 0.96, 0.9189, 0.9074, and 0.8865, respectively, and the average alarm delay is about 15 seconds.