The Research And Implementation Of Self-tuning Method Based On Software Behavior Monitoring

In the dynamic, open and personalized web platform, the changes of environment and user requirements affect the software easily. In order to maintain the normal operations of the software, management and maintenance of software systems become heavy burdens increasingly. So the self-tuning technology was proposed in this thesis to give the ability for self-adaptation for software systems. The software can perceive the changes of environment. According to these changes, the software can take adaptive action to adapt to changes in resources and user needs to maintain the system functionality or services performance at a satisfactory level.First of all, the theoretical basis of software self-tuning application is discussed in this thesis. Then in the use of reflective system, a software self-tuning model based on software behavior monitoring is created. This model can improve the flexibility of the software systems. So the systems can adapt to environmental changes and changes in demand better.Secondly, software behavior monitoring technology based on software behavior tree is studied. The technology can determine and predict the software running more accurately. So that software could adapt a more appropriate self-tuning strategy. On the basis of researches about methods of self-tuning in two language environment of VC and JAVA, a self-tuning algorithm based on properties is proposed in this thesis. This algorithm uses the method of tuning component property to make the software adapt to changes of external environment to achieve the purpose of software self-tuning.Finally, the overall program of the self-tuning system based on software behavior monitoring is designed, and each module of the system is designed in details. Then, the function of each part of the system was tested. The experiments have verified the feasibility of the system. The research priorities for the next step are presented at the same time.