QoS Aggregation in Web Service Compositions

TLDR

Web service composition typically uses non‑functional characteristics to select services, and this aggregation relies on abstract composition patterns that model structural elements such as parallel paths, sequences, or loops. The study aims to develop a mechanism that aggregates individual service QoS to determine the overall QoS of a composition, extend patterns to account for service dependencies, and apply this pattern‑based aggregation during runtime monitoring. The mechanism aggregates QoS by applying pattern‑based calculations, incorporating dependency‑aware extensions, and using runtime monitoring data to refine the aggregation for greater accuracy. Aggregated QoS enables verification of whether a set of services meets the overall composition’s QoS requirements.

Abstract

For the composition of Web services non-functional characteristics are commonly considered criteria for finding and selecting available services. Our work focuses on a mechanism that determines the overall quality-of-service (QoS) of a composition by aggregating the QoS of the individual services. With aggregated QoS it can be verified whether a set of services satisfies the QoS requirements for the whole composition or not. The aggregation performed builds upon abstract composition patterns, which model basic structural elements of a composition like parallel paths, a sequence, or a looped execution. In this work we extend existing composition patterns with the abilities to consider dependencies between services. Furthermore we introduce, how to use the pattern-based aggregation in the monitoring process during run-time. We explain how the data derived from the monitoring process can be used to calculate a more accurate aggregation of QoS for the composition.