Design And Implementation Of Linux Server Performance Monitoring System

In recent years,with the rise of the computer industry and the explosive growth of Internet users,it is necessary to ensure data security and reliability while meeting the needs of the market and users,The traditional monitoring system fails to analyze and utilize historical server performance data.It provides early warning after the server fails.Although the early warning function can be realized,the server has already failed during the early warning,which seriously affects the operation of the process.Referring to the mode of server monitoring system at home and abroad,this thesis designs and implements a Linux server performance monitoring system,which realizes the early prediction and early warning of faults,not only improves the utilization of server resources,but also ensures the normal operation of the process.The specific work is as follows:(1)According to the demand analysis,the system is designed into four modules: system management,collection task management,server performance data monitoring and server performance data prediction.The system management mainly uses Layui front-end framework to call ECharts library for display;the collection task management is mainly realized by the Quartz framework;the server performance data monitoring is realized by reading the corresponding data of the /proc file under Linux through Java;the Server performance data prediction is predicted by the improved ARIMA-BP combined model.(2)The choice of model,because the server performance data contains linear and non-linear data,and the ARIMA model only has a higher prediction accuracy for linear data.Therefore,this thesis proposes an improvement based on the traditional ARIMA model,which uses the average value method to establish the prediction model,and realizes the error correction by setting the minimum error precision,This model improves the nonlinear prediction data of traditional models.The average absolute error,mean square error,and root mean square error are used to evaluate the experimental results.A combined model is proposed according to the improved ARIMA model idea.A BP neural network model for nonlinear data prediction is introduced,The residual of the improved ARIMA model is taken as the input value of BP neural network,and the prediction result of the combined model is the improved ARIMA model plus the prediction residual output value of BP neural network,which further improves the accuracy of the improved ARIMA model.(3)Design and implement the four modules of the system.The system management module mainly manages the basic information of system users and system roles;the collection task management module mainly collects the server performance data regularly;The server performance data monitoring module performs real-time monitoring of CPU usage,memory usage,and network bandwidth usage and realizes pop-up warning functions;the server performance data prediction module realizes the daily prediction and weekly prediction of three kinds of data and SMS warning function.