QoS aggregation for web service composition using workflow patterns

TLDR

Web service composition depends primarily on matching semantic descriptions and on the quality of service (QoS) attributes. The paper introduces a mechanism that computes the overall QoS of a composition by aggregating the QoS dimensions of its constituent services. This mechanism aggregates QoS across services using abstract composition patterns—sequence, loop, and parallel—derived from Van der Aalst’s workflow patterns, producing an aggregation schema that mirrors workflow structures and supports multiple QoS dimensions. The resulting approach allows verification that a selected set of services satisfies the overall QoS requirements of the composition.

Abstract

Contributions in the field of Web services have identified that (a) finding matches between semantic descriptions of advertised and requested services and (b) nonfunctional characteristics - the quality of service (QoS) - are the most crucial criteria for composition of Web services. A mechanism is introduced that determines the QoS of a Web service composition by aggregating the QoS dimensions of the individual services. This allows to verify whether a set of services selected for composition satisfies the QoS requirements for the whole composition. The aggregation performed builds upon abstract composition patterns, which represent basic structural elements of a composition, like sequence, loop, or parallel execution. This work focusses on workflow management environments. We define composition patterns that are derived from Van der Aalst's et al. comprehensive collection of workflow patterns. The resulting aggregation schema supports the same structural elements as found in workflows. Furthermore, the aggregation of several QoS dimensions is discussed.