Modeling Users' Visiting Behaviors for Web Load Testing by Continuous Time Markov Chain

Abstract

Virtual users with high quality are the preconditions to ensure the effect of load testing for Web applications. The existed tools for load testing usually generate virtual users with randomly choosing user sessions, manually generating user sessions or mining Log files, which causing such problems as non-real workload, subjectivity or difficult to update. Therefore we set each virtual user with a corresponding configure file, and these files determine the visiting paths, visiting moments and stay time of virtual users based on the Continuous Time Markov Chain. So we firstly finish the pretreatment for Log files, then construct the user visiting model, and next generate the virtual users, lastly carry out the load testing. In this way, we can obtain more reliable results for Web application load testing than the existed methods.