Research On The Tendency Of Software Aging For The Web Technology Developed Software

With the development of the web technologies and the increment of users’ demand, thenumber of software based on web technologies is growing and the structure becomes morecomplicated, it brings convenience to users, yet at the same time it is difficult to maintain thesoftware. In a long-running software system the unknown legacy defects of software will beaccumulated to produce phenomena such as the leaking of memory, the no-releasing offile-locking, the storage space fragment, the accumulation of errors and so on. These will leadto performance degradation eventually a sudden crash of software system. The phenomenonof software aging reduces the software system’s reliability. In order to reduce the harm ofsoftware aging, we need to research the software aging for the web technology developedsoftware and take timely measures to improve software performance.By analyzing the existed forecasting methods, this thesis argued that the time seriesanalysis method is recognized as the basic method. The time series of software performanceparameters were the main research object of forecasting method. This method usedmathematical approaches to analysis the future trendy objectively.In this thesis, we made adeep research on the time series analysis method, analyzed the existed problem and improvedthe method. For the issue of determining order method no longer change, an adaptive methodwas used to choose the given order method. For the issue of deviations in the time sequenceprediction, the time sequence results by the state interval would be corrected when theMarkov determine the prediction time, that is, time series and Markov would combinetogether to predict the tendency of software aging.In order to prove the forecasting model can predict software aging better, the experimentverified the effectiveness of the model. By analyzing a variety of performance parameters,this thesis regarded the used memory, average load and response time as the key parametersfor performance recession. This thesis collected the performance parameters of HanbIntegrated Query System running based on web development technologies. Matlab, Eclipseand SQL Server2005were the major tools to achieve the prediction of the forecast of thetime series model and used Markov to correct the results.