Abstract
Enterprise information systems consist of interrelated Intranet-based and Internet-based applications, thus the Web platform serves well as the middleware for their development. Even though, in many cases, discrete applications operate efficiently, the overall system performance is less than expected. A potential cause is that, configuration issues although interrelated are solved in isolation. The architecture and performance characteristics of the underlying network affect application configuration. Thus, the symbiotic relationship between application and network architecture must also be explored. We argue that, in order to effectively configuring and evaluating enterprise information systems, network infrastructure restrictions must be incorporated within application configuration process and vise versa. An effective four-stage methodology is proposed for this purpose. It is important to note the significance of a consistent model for the representation of system entities throughout all stages. UML-like notation is used for system representation. Alternative views of the system emphasizing specific configuration stages are offered through UML package representation. Since the modeling scheme is extendable, the adaptation of UML constructs simplifies the process of extending or customizing the model. A case study where the proposed methodology is used for the configuration a large-scale medical information system and the experience obtained are also presented.