Simulating e-Commerce Client-Server Interaction for Capacity Planning

Abstract: Contemporary ways of doing business are heavily dependent on the e-Commerce/e-Business paradigm. The highest priority of an e-Commerce Web site's management is to assure pertinent Quality-of-Service (QoS) levels of their Web services continually, in order to keep the potential e-Customers satisfied. Otherwise, it faces an immense possibility of losing both e-Customers and revenues, along with a big possibility of gaining bad reputation due to either poor performance or unavailability of the e-Commerce Web site. In order to avoid numerous unpleasant consequences, by designing and implementing e-Commerce Web sites that will always meet e-Customer's high expectations, a relevant performance models have to be utilized to obtain appropriate performance metrics. A continuous assessment of current performances and especially predicting future needs are the subjects of capacity planning methodologies. Within this paper, such a predictive model has been described and evaluated by using discrete-event simulation of both the client-side and server-side processes involved. In addition, the paper discusses the performance metrics obtained as a function of the intensity and quality of the workload parameters.