Web应用的统计测试与可靠性度量体系研究

Because of the complex Web application environment, the reasons result in Web page failure are various, not only the software internal failures, but also the user usage and network environment. So the traditional measuring methods of software testing and reliability evaluating are facing many new challenges. Generally the traditional software testing and evaluating can be evaluated quantitatively by using reliability model, but most of the models are based on software internal failures, ignoring the user interactive behaviors with the system. Therefore, the traditional method based on failure data to test and measure the Web application system reliability can not be good to meet the demand. This project gives a new statistic testing and reliability evaluation system based on web server logs. The system collects workload of web application system from web log, and regards workload as users’ behavior. Then workload can be the input of Markov chain based usage model, ODC (Orthogonal Defect Classification) analysis method and classical reliability models. And then the system become a whole system, include testing, tracking defects and reliability evaluating.

Web应用环境复杂，导致Web页面失效的因素多样，包括软件内部故障、用户使用和网络环境等诸多因素，给传统软件测试和可靠性度量方法带来新的挑战。传统软件测试与可靠性评估一般是按模型进行定量评估，而大多数模型又都是基于软件内部故障进行建模，忽略了用户与站点的交互行为，因此，传统基于故障测量的软件测试和可靠性模型不能很好的满足Web应用系统测试和度量的需求。本项目提出一种结合Web日志的统计测试和可靠性度量体系，从Web日志中提取Web应用的工作量，以此来模拟用户与应用的交互，提出一种基于马尔可夫链的序列生成和分析的统计测试方法、ODC正交缺陷分类分析法和可靠性度量模型，形成一整套针对Web应用的测试、缺陷跟踪、可靠性度量的完整体系。

Web应用环境复杂，导致Web页面失效的因素多样，包括软件内部故障、用户使用和网络环境等诸多因素，给传统软件测试和可靠性度量方法带来新的挑战。传统软件测试与可靠性评估一般是按模型进行定量评估，而大多数模型又都是基于软件内部故障进行建模，忽略了Web应用相关的情境信息，因此，传统基于故障测量的软件测试和可靠性模型不能很好地满足Web应用系统测试和度量的需求。本项目提出一种结合Web应用开发、运行和维护数据的统计测试和可靠性度量体系，提取Web应用的易错性、易变性和工作量等典型的质量指标，提出基于马尔可夫链的序列生成和分析的统计测试方法和代码异味感知的可靠性度量模型，形成一整套针对Web应用的测试、缺陷跟踪、可靠性度量的完整体系，并指导Web应用代码的重构。