A Model-Based Performance Testing Toolset for Web Applications

Effective performance testing techniques are essential for understanding whether a Web-based application will meet its performance objectives when deployed in the real world. The workload of such an application has to be characterized in terms of sessions; a session being a sequence of inter-dependent requests submitted by a single user. Dependencies arise because some requests depend on the responses of earlier requests in a session. To exercise application functions in a representative manner, these dependencies should be reflected in the synthetic workloads used to test Web-based applications. The need to handle these dependencies makes performance testing a time consuming process. Specifically, since the nature of these dependencies varies across systems considerable effort needs to be dedicated towards tailoring a set of test tools that will work for a given system. Furthermore, the traditionally followed approach of manually developing test scripts that handle dependencies makes it difficult to achieve fine-grained control over a synthetic workloadʹs characteristics. In this paper, we propose a new model-based approach to address these issues. Our approach uses an application model that captures the dependencies for a Web-based application system under study. The application model provides an indirection that allows a common set of workload generation tools to be used for testing different applications. Consequently, less effort is needed for developing and maintaining the testing tools and more effort can be dedicated towards the performance testing process. Furthermore, the application model can be used to obtain a large set of valid request sequences representing how users typically interact with the application. This feature facilitates techniques that can manipulate the set of sequences to achieve automated control over the characteristics of the synthetic workloads used in a testing exercise. The utility of the toolset is demonstrated by applying it to model two different Web applications and through a study that systematically investigates the impact of various workload characteristics on the performance of an e-commerce application.