Modelling Website Infrastructure using B-Node Theory
Faculty of Computing, Health and Science
School of Computer and Security Science
A large volume of business is conducted via the Internet (Schneider and Perry 2000). However, this has resulted in increased transaction delay times as systems and computer networks become overloaded (Devlin, Gray. J et aL 1999). Surveys and studies indicate that slow downloading time is the most often cited reason that an online customer leaves a site and searches for another vendor's site (Bakos 1998). According to Shklar: "Sites have been concentrating on the right content". Now, more of them specially e-commerce sites realize that performance is crucial in attracting and retaining online customers."(Shklar 1998). The performance of an Internet site is dependant not only upon the behavior of end users using that site but also the performance of the technologies employed. Currently there are a number of different models for defining e-business web sites performance, such as the 'business', 'functional', 'customer', and 'resource' models (Menasce, Virgilio et .aL 2000). The Customer Behavior Modeling Graph (Inverardi & Wolf, 1995) (Union, 1996) and Client /Server Interaction Diagrams (CSIDs) (Stohr and Kim 1998) are techniques that can be used to capture the navigation patterns of customers during site visits and hence obtain quantitative information on workloads. However, although these models attempt to predict user behavior, they do not provide information about the actual load on the systems running on such sites. In the final analysis user load must be translated to hardware requirements thereby allowing performance bottlenecks to be identified. However, such problems associated with infrastructure design are non-trivial. According to Fenik "being able to manage hit storms on commerce sites requires more than just buying more plumbing." (Fenik 1998). In order to predict the workload characteristics of e-commerce sites, effective modelling needs to be undertaken to determine key bottlenecks within the system. It is therefore necessary to investigate different types of models which could be used to model the infrastructure of e-commerce web sites.