Performance Analysis of Web Applications Based on User Navigation

This paper proposes a method to conduct performance eanalysis of web applications. The behavior model is firstly built from log file after user navigation, then an extended state diagram is extracted from this log file, finally multiple Markov model is cooperated to this state diagram and the performance analysis can be obtained from the Markov model. Five indexes are used to measure the performance and they are: service response time, service path length, service utilization, service implementation rate and access error rate. Our performance analysis result will provide a suggestion to improve the design of web applications and optimize the services. A case study of Zhejiang Chess web site has been used to demonstrate the advantage of our method.