Research And Implimentation Of Applications Run Trend Forecasting And Analysis Subsystem In APM System

Application performance management system through the enterprise's key applications to monitor and analyze to ensure the normal operation of the system,allowing users to get reliable services,reduce operating costs and improve business efficiency.Predicting future system operating conditions and anomalous alarms are two important requirements for application performance management.The traditional time series prediction technology has the disadvantage that the analysis process needs too many human interventions and the prediction accuracy is not high.Compared with the traditional method,the deep learning technology has the characteristics of self-adaptability.This paper applies the deep learning technology to the application performance prediction and abnormal mining in application performance management.Based on the neural network technology,the paper designs a prediction model of typical resource consumption forecasting system.The neural network model is used to mine the contextual relationships of various data during the operation of the application,and a regression prediction model applying historical resource consumption and future typical resource consumption is established.Analyzing the neural network's prediction output provides the system with capacity management recommendations for some time to come.In the aspect of system abnormal change mining,the relative entropy distance model for predicting residuals was designed.In this paper,firstly,the background knowledge and related technologies are introduced,and then the overall structure of the application performance management system is linked up.The position and work scenarios of the application running situation prediction and analysis subsystem in the whole system are described,and the specific requirements of the application performance management system are analyzed and the specific function points of subsystems are summed up.In the latter part of the algorithm model design,two key analysis models of time series prediction and application abnormal change detection are given.In the part of subsystem design and implementation,the subsystem structure,key processes,internal and external interfaces,data structure definition and system class diagram design are introduced.According to the fifth chapter of the design content,completed the prototype development of the subsystem.In the sixth chapter,the test cases are designed for the function points of subsystems and the verification is carried out.The test results are basically in line with the expectation.In the sixth chapter,the test cases are designed for the function points of subsystems and the verification is carried out.The test results are basically in line with the expectation.