An Automated Client-side Monitoring Technique For Web Applications

Web applications have become more and more popular today,and a new generation of Web applications using Ajax,like Amazon and Gmail,includes a significant client-side JavaScript component running in the browsers.With the diversity of client-side environments and browser configurations,it is difficult for us to completely test or debug the client-side code of Web applications in-house.There are tools that facilitate functional testing on various browsers,but they can not mimic all of the possible client-side environments as both Web applications and client-side environment are complex.In modern Web browsers,the client-side code can interact with numerous Web services to get more data and to update itself, which in turn affect the behavior of the client in unforeseen ways.Existing techniques allows for collecting some client-side runtime information,such as the URL of the page where an error occurred and error messages.This information,however, provides only limited support for debugging problems occurring in the client-side code.In these situations,monitoring the client-side code allows for gathering valuable runtime information about its behavior.In this paper, a new technique for monitoring and detecting failures in client-side code is proposed.Using a proxy server, a client agent is transparently injected into the Web application's code,and then the client agent returns the collected monitoring data to proxy server, then a proxy-side component analyzes the monitoring data and can even control the client agent to take corrective actions.This paper also presents a preliminary evaluation of the technique about its possible application scenarios,its efficiency and its effectiveness.The result indicates that it is a flexible,effective monitoring and testing technique with rich functionalities for Web applications.