Performance Analysis of E-Commerce Site:Modeling of Customer Behavior by Multi-Class Queuing Networks

郑小林,陈德人
DOI: https://doi.org/10.3785/j.issn.1008-973x.2004.09.004
2004-01-01
Abstract:A method of workload characterization of e-commerce site by modelling customer behavior was presented. First, a transition model of customer stations was used to characterize the business functions that a customer performed in a session and the relations between these functions. An open multi-class queuing network was used to model the customer behavior in the e-commerce site system, and the bottleneck server of performance was found out. Furthermore, a closed multi-class queuing network was used to model this server, and helped to find the bottleneck components in this server. A case study indicated that some of the important workload parameters could be calculated easily. It shows that this method can find the bottleneck of e-commerce site quickly, and can instruct the performance analysis and capacity planning for e-commerce site.
What problem does this paper attempt to address?