Aggregate QoS Estimation of Service Compositions - An Analysis of Pattern-Oriented Approaches

For service-oriented systems the estimation of QoS is an important factor to determine whether a prospective service composition will meet the expectations of its final users. Many approaches to QoS estimation of service compositions have been proposed. However, not all of them allow the QoS of composition be estimated a priori and systematically from the QoS of its individual services. This is an important goal for many composition-based development approaches. This paper is a first effort to analyze QoS estimation approaches aligned to this goal. We revise approaches that use patterns as composition mechanisms as they enable the QoS of a service composition be estimated, a priori and systematically, by aggregating the QoS information of its constituent services. The result of this analysis offers insight in the main patterns and quality attributes addressed by these approaches as well as the adopted aggregation criteria. This will help us to achieve our ultimate goals of using these approaches on public services, assessing them for practicality, identifying gaps between theory and practice and possibilities for future work.