Research On The Prediction Of Web Application Software Aging Based On User Action

Software aging phenomenon is commonly found in Long-pasting operation of the software, which is the serious threat to the reliability of software and the computer system and become a major bottleneck restricting the development of software reliability. How to solve the software aging had become the key point which the academic field pays attention. At present, Software rejuvenation is used to solve this problem. The software rejuvenation has certain regeneration cost and therefore forecasting the software aging is of great significance to determine the regeneration time, and thus to reduce the regeneration cost. The traditional methods of predicting the software aging is suboptimal in the case of the big changes of system load and the large amount of concurrent. When we measure the software aging from system performance point of view, it is difficult to ensure the result accuracy and highly real-time since the index concepts of measuring system performance is obscure and is difficult to implement.As the user action with a certain regularity which can be followed and controlled, this thesis proposes a prediction method of Web application software aging based on user action, which take user action, the load of Web application-as the research object and predict the Web application software aging by predicting it. Firstly, this thesis proposes the whole process of prediction method to Web application software aging, and then introduces several key points and the research ideas in this process in detail. Then it presents the user action and the predict method proposed. Based on the characteristics of certain laws and patterns of user action, this thesis proposes the prediction method based on FUAP-Tree, which zip the user’s access action into a FUAP-Tree for storing the Web log firstly and predict user behavior by pattern matching through traverse the related sub-tree of FUAP-Tree. The load of Web application software can be obtained under the circumstances that the user’s access action is known in the future, and therefore we can predict the Web application software aging in this moment.Experimental results prove that the feasibility of the prediction method of Web application software aging and the certain effectiveness and accuracy.