Study And Implementation On Performance Tuning Based On J2EE System

In recent years, Enterprise Application Systems had attracted widely attention and been put into practice by more and more enterprises. With the rapid growth in Application area, Enterprises have required high usability, reliability and extensibility, but with less expense.Judging from J2EE development process and core practice technology, we find that J2EE multilayer application frame increases software's importance and breaks problem's complexity as well as makes codes numberless and the control among layers complex, which always bring in inconsistencies between the consistency of performance, load, efficiency, data and extensibility of groupware, with the developers easily falling into performance trap or producing unqualified products. The main result is when accessing clients or data throughput increase; the request will be delayed or even be failed. There are so many papers which put on focus on J2EE application optimize tuning. They just analyze and discuss on some aspects of the system's trait. In reality, most of the strategies are basing on the official documents, which are offered by the web application servers'manufacturers. The staffs who are tuning the web servers need to analyze the documents to find the most important parameters in adjusting the server. The traditional tuning way seems so difficult for a layman to fully understand the meanings of different parameters in a short time. The traditional tuning way also lacks of regular methods to find the correct parameters. It almost depends on the experiences of web tuning staffs.In this thesis, application server's low performance was analyzed and J2EE performance tuning was band with the attributes reduction theory in decision system of rough fuzzy sets. Parameters modules which inspirations came from analyzing the low efficiency of server were also introduced. Binding parameters models to the attributes reduction tuning were introduced by this thesis. We found the key parameters from practical using and gave them specially tuning for swift performance optimize. That way was different from the traditional tuning way. It came from experiments which could pick up the key parameters with the mathematical tools. Proved by the practices, the way indeed took great effects in tuning application servers and had the characteristics of easily mastering and fast tuning.According to the practice situation of J2EE application system's development and practice using, with an eye to Pattern design, groupware reusing, coding optimizing and application deploy, a scheme which was referring to different aspects of the performance for the system optimizing was proposed in the thesis.