Evaluating service identification with design metrics on business process decomposition

Abstract

One of the key activities to construct a successful Service-Oriented Architecture (SOA) is the identification of services with the right level of abstraction. Most of existing SOA design methodologies advocate identifying services from the top-down decomposition of business processes. However, the identification quality in these methodologies heavily depends on the expertise and experience of individual designers. The ability to quantitatively evaluate service identification is absolutely needed. In this paper, we propose an approach that supports such evaluation by applying the measurement technology to the service-based business process decomposition. A model to capture related architectural elements with their relationships is presented. A set of design metrics are proposed for measuring various features of identified services in the service portfolio, including service granularity, coupling, cohesion, and business entity convergence. To apply the approach, a prototypical measurement tool for service identification is developed. An automotive work order scenario is used as an illustration example to explain our approach and demonstrate its effectiveness. © 2009 IEEE.