The Design And Implementation Of JVM Performance Monitor System Based On Fabric

With the rapid development of network, more and more companies are used to using J2EE to develop web applications. Meanwhile more and more users are beginning to realize the importance of product non-functional performance. As a result, people paid more attendtion to application performance, espeacially those who delevelop and test applications. In eBay, to ensure the product is running normally, testers often need to log on a remote host to see the Java application performance by using command-line tools, which is frequent and tedious. Therefor, we designed and implemented a Java application performance monitoring system based on web to help user.The System will log on a remote host to collect performance data automatically and periodically for a specified java application after it received a performance monitor request from user and save the data to database. User can see the performance reports from web page. This thesis describes how to design and implement the system. System is divided into two subsystems:monitor system and console system. Console subsystem is a web application and focused on providing user-friendly Ul for user while monitor subsystem is a web service, which is responsible for monitoring java application performance. Based on the web pages provided by console subsystem, user can easily add, delete, edit, start and stop minotoring tasks for remote java application. What’s more, in console subsystem the performance result will be presented in the form of charts and tables which will be more intuitive and vivid.According to the functional module division principle, console subystem is divided into task management module and performance reports module. Monitor subsystem provides interface by RESTful API and includes task management module, performance-monitoring module, warning module and email notification module. The performance-monitoring module is critical in the system, which is responsible for data collection and persistent storage. Warning module sends warning to user when there is abnormal performance. Email notification module is in charge of sending emails to user according to user preference when the performance task completes or goes wrong or an warning rasies. Console subsystem does not process business logic and it implements the function by calling web service provided by monitor subsystem. Monitor subsystem takes charge of implementing business logic for system.System is developed by J2ee based on spring boot framework and uses MongoDB. However, remote data collection program is implemented by Python based on Fabric. AJAX, RESTful web service, multithreading and more technology is used when developing.