Approaches To The Prediction And Enhancement Of Test Scripts For Memory Bloat In Java Web Applications

Recent years,with the rapid development of web techniques,Web applications have been widely used in many areas.However,the popularity of web applications also introduces more problems,one of which is memory bloat.Memory bloat is a common fault in the Web applications and middleware.With the long time running of the application,software memory consumption is increasing and then memory bloat will affect the system performance.For Web applications,obvious memory bloat problems were exposed after a long running time many times.The traditional testing methods are lack of guidance and waste a lot of resources and time.And lacking the ability of finding errors in the limited time.We proposed Java Web application bloat script prioritization,optimization,and prediction approaches to help resolve Java Web memory bloat problem.Test script optimization is a process of cutting the unused statements of test script to accelerate the script execution process.Test script enhance is a process of add loop or concurrent statements to test script.It could expose the memory bloat obviously by running the enhanced script.(1)Firstly,we proposed an approach to predict and ranking the scripts.We collected the memory bloat related data while running the Web application,and transform these data to a matrix,and the based on the matrix,we train a memory bloat prediction model.At the last,we apply the model to the measured Web applications and ranking their scripts by the memory bloat risk.(2)Then,we proposed an approach to optimize and enhance scripts.We optimize the scripts by remove the statements which are irrelevant with the w eb execution.We enhance the scripts by add loop or concurrent statements to the scripts,so that we could get enhanced performance scripts.(3)Finally,we developed a memory bloat test tool to support the above two approaches.Experimental evaluation on several Web applications shows that our method is effective.