Software Evolution Based Performance Evaluation And Analysis

The rapid development of computer technology prompts users to have higher expectations and requirements for the functions and performance of all kinds of software,and the cost of software development and maintenance is also increasing.With the changes of software functional requirements and the dynamic changes in the environment,the evolution of software inevitably occurs.How to ensure that the performance indicators of software is not reduced during the process of evolution is an important issue to keep software quality from a non-functional perspective.At present,software performance testing is an effective way to measure performance of software.It can measure various performance indicators of a system by simulating various load conditions and evaluate software quality from a non-functional perspective.However,there is still a lack of methods and tools to measure and evaluate the performance changes in the process of software evolution.Therefore,how to measure and evaluate the performance changes of software during evolution and find the root cause of a significant decline in performance is the focus of this thesis.Based on the investigation of software evolution,software performance evaluation methods and performance monitoring tools,the existing performance evaluation methods are improved,and the performance evaluation and analysis methods based on software evolution and the corresponding performance evaluation tool are put forward.The details are as follows:(I)through the research on the software performance indicators,combined with the characteristics of Java software,the performance indicators to be evaluated are selected,these indicators include processing time,CPU occupancy and memory occupancy;(2)put forward the software performance evaluation and analysis method based on software evolution,through the performance evaluation and comparison of multiple versions of Java software,we determine the versions which have great performance changes and then analyze and locate causes of performance changes through code differences and performance differences between two adjacent versions of software;(3)on this basis,we design and implement a performance evaluation tool based on software evolution for Java applications running on Windows,and verify the validity and correctness of the method and tool through experiments on open source software.