Service-oriented architectures: Modeling the selection of services and platforms

Abstract

The Service-Oriented Architecture (SOA) paradigm promises to facilitate the integration of software services provided by different vendors and thus enables users to benefit from Best-of-Breed solutions. In order to support software architects we present the Multilayer Standardization Problem (MSP) to analyze the trade-off between possibly enhanced utility versus higher assembling costs of Best-of-Breed SOA solutions. We implemented a software prototype to support decision makers during the data input and the subsequent analysis of the solution's robustness. The MSP for the SOA-case is formulated as a linear 0-1 optimization model and extends the established Standardization Problem (SP) by modeling the user preferences and considering varying granularity as well as integration relationships in addition to communication relationships. These characteristics are common to numerous systems - thus the general MSP can serve as a basis for further research in this field.