A Progressive Real-Time System Performance Defect Diagnosis Framework Based on Java Virtual Machine Run-time Data

Effective diagnosis of real-time system performance defects is of great significance in practical scenarios. Traditional performance defect detection methods are limited in localization and time-cost, which restrict the use of real applications. To address the above problem, this paper proposes a progressive real-time system performance defect diagnosis framework based on Java Virtual Machine (JVM) run-time data. This framework first conducts the fast diagnosis to derive the preliminary performance evaluation results based on the status data from the operating system during the running period of a real-time system. secondly, conducts the deep diagnosis to analyze the cause of defects in depth based on the stack data from the JVM run-time area. Finally, utilizes a deep-learning method to capture the potential relationships of stack data in the JVM run-time area and learns the disposal policies for the detection task. In addition, we instantiate the framework to a real-time scenario in the field of civil aviation e-commerce tickets and design the core data structure and functions list of the prototype diagnostic system for the sales of civil aviation e-commerce tickets.