An architecture-based QoS prediction model for web service composition

Abstract

SOA (Service Oriented Architecture) has become a promising paradigm by making developers to easily develop software system by just integrating existing web services. To maximize such advantages of SOA, web service composition is very important role in binding dynamically individual web services, which are distributed across the internet, at an execution time. However, since participant web services consisting of composite web services are distributed and loosely coupled at runtime, it is very important to evaluate whether QoS (Quality of Service) of composite web services meets non-functional requirements driven from service consumers, in order to prevent web service composition from re-occurring. For this, five types of QoS prediction models such as on the basis of workflow, Markov chain, aggregation function, simulation, and failure-aware have been proposed. However, existing QoS prediction models have some common limitations as following: Firstly, existing models do not take interactions among web services into consideration, even though it is important factor for QoS prediction since web services are distributed across the internet. Secondly, those models do not take coordination models (such as orchestration or choreography) of composite web services into consideration, even though the interactions behaviors are different according to coordination models. Thirdly, those models focus on specific one or two QoS attributes, rather than providing various QoS attributes for QoS requirements. To address above three major limitations, we propose an architecture-based QoS prediction model which consists of quantitative and qualitative QoS prediction models for composite web services. Quantitative prediction model is to calculate quantitatively QoS of composite web services by using extended aggregation function. On the basis of software architecture transformed from business process workflow, the interaction data are obtained from WSIM (Web Service I...