| With the development of computer software technology, more and more enterprises solve their own core issues by software. However, long-lasting operation in the software process, system resources will gradually exhaust due to the accumulation of the running errors, it results in system performance continues to decline and even hangs off, this phenomenon is called software aging.Software aging, which is a serious threat to the reliability of software, at present, commonly using software rejuvenation technology to solve this problem. The forecasting of software aging trend is a key to software rejuvenation, which is used to determine when to regenerate.The traditional time-based and measurement-based are limited in dealing with the large number of visitors and the large amount of concurrents in Web system,therefore,this paper presents a method-prediction of Web system aging trend oriented to user access intention.The method researches this problem based on user load,predicts software aging trend based on user intention.Firstly,this paper proposes the whole process, and then detailed analyzes several key points.Then focus on how to predict the user intention, there are two key algorithms, sequential pattern mining algorithm and pattern matching algorithm.This paper improves the traditional PrefixSpan and KMP algorithms to achieve the purpose.Finally, estimating the amount of access from the user intention,combined with the amount of damage to predict the trend of aging of Web server.In Web system, user access is not random, but with a certain pattern, using the frequent patterns which are from data mining to predict the intention of users.Experimental results show that this prediction is accurate and efficient.Using the prediction result to master the load of Web server in the future period of time, consequently, to predict the ageing trend.Through the comparison the prediction result with the simulation result to prove the feasibility of this method. |
PDF Full Text Request