Abstract
Before placing a web based application in the real servers of the real world with millions of users and going to serve them, it is needed to examine the functionality of the application which is under real work load stress to evaluate server and application for final conditions. This paper is focused on modeling of the user behavior by using data mining methods and extracting different types of application clusters and creating profiles of user's behavior for each cluster. Also user think time is considered as the principal component of work load and presents a mathematical model of user think time. Related to the selected web application and finally by using the distribution function of user think time and written macros for the web browser, it also creates the work load of types of users and sends it toward the web servers and examines web server status. This article unlike some studies that attempt to model web server and characterizing it's components, has simulated the web server responsibility and ability by using real traffic to prepare better solutions if hardware or their applications codes are not suitable. Unlike the related works that validate a special system and characterize parameters of its load, our recommendation is useful for all types of servers and web applications. Due to proposed model results, it's possible to prefer suitable recommendation for programmers and the relevant site management.