Collaborative and corroborative semantic web service monitoring

Abstract

Service Oriented Computing has emerged as a promising computing paradigm for Internet-scale distributed applications built around services as primary building blocks. Such Internet-scale computing relies heavily on the connectivity infrastructure because existing business functionalities could be made accessible through this infrastructure. Furthermore, the services, as building blocks of these Internet-scale applications, may be developed and managed autonomously by enterprises. Different organizations may provide similar functionalities and it therefore becomes a matter of differentiation to support negotiable quality of service characteristics such as response time, cost, throughput, among others. Service consumers consider these non-functional characteristics of the published services when selecting which service to use; so there is an economic aspect to quality of service (QoS). The negotiated QoS characteristics often involve penalties, making it absolutely critical for the service provider to detect and correct any deviations from the agreedupon service behaviour. The need to detect potential failures has resulted in several research activities dedicated to service monitoring, specifically, Semantic Web Service Monitoring. Monitoring entails detecting and signaling whether the participating services behave consistently with the expected functionality and non-functional service properties. Monitoring may further entail enacting corrective mechanisms when there are failures or breaches to agreements by contracted services. Many approaches for both the service-consumer-side and service-provider-side monitoring have been proposed, including the use of dedicated monitoring infrastructure. Contemporary monitoring approaches rely on the service performance data being collected separately by the service provider for service-side monitoring and the consumer for client-side monitoring. As of this writing and according to our knowledge, no monitoring approach had been developed that facilitates the collaborative and corroborative exchange of monitoring information by both the service consumer and the service provider. The exchange of monitoring information is significant because: (1) the service consumer and the service provider may have different perspectives of the same QoS parameter and collaboration can help develop consensus, (2) enables the support for a flexible quality-based pricing model, which is a potential competitive advantage for service providers, (3) it has a built-in self-checking mechanism so that there is no need for costly incentive schemes to encourage honest reporting, (4) there is no need for costly dedicated infrastructure for monitoring by surveillance because service consumer and service provider exchange their context-dependent monitoring information directly. Having made a case for a collaborative and corroborative monitoring approach, this study hypothesizes that collaborative and corroborative monitoring of semantic web services is a viable alternative approach to client-side, service-side and 3rd-party/dedicated monitoring infrastructure. Therefore, the main objective of this study is to investigate, design, develop, and evaluate a collaborative and corroborative monitoring technique for Semantic Web Services. Design Science Research (DSR) methodology is adopted as it is particularly suitable for research studies that, like this study, aim to design an artefact. The significance of this work is underpinned by the critical importance of effective monitoring for the practical application of Service Oriented Computing technologies, specifically the Semantic Web Services. This study is critical because it purports to introduce a novel and practical approach to Semantic Web Service monitoring that is corroborative and collaborative between a service consumer and a service provider. In pursuance of the collaborative and corroborative monitoring, the study developed a Generalized Response Time Metric (GRTM), a consensus-based generalized metric for response time QoS parameter. The study further developed a technology-agnostic Monitoring Information eXchange (MIX) protocol to facilitate the exchange of performance data which is described in terms of the GRTM. The study also found, and demonstrated, that the support for, and implementation of collaborative and corroborative monitoring for Semantic Web Services is technically feasible. This is made possible by implementing a monitoring information exchange mechanism based on the MIX protocol. The proposed solution has been evaluated in terms of its technical feasibility. There were no other implementations of collaborative or corroborative monitoring of Semantic Web Services at the time of the execution of the research and consequently, there was no need for comparative analysis. The study demonstrated the technical feasibility of the proposed collaborative and corroborative monitoring technique for Semantic Web Service. This was achieved through a reference implementation of the MIX protocol. Finally, the study makes three recommendations for consideration as future research work and these are based on the identified limitations of the study. Firstly, developing the correct syntax for the semantic description of the mathematical expressions of all the parameters that characterize the performance of a Semantic Web Service such as latency, response time, throughput, and error-rate. An expressive ontology model for Semantic Web Service performance needs to include the logical expressions of these performance parameters; the study focused only on response time. Secondly, a Semantic Web Service Monitoring tool based on the MIX protocol should be developed further to provide extensible interfaces that software engineers can implement to support MIX functionality. The third recommendation is for software engineers and researchers to explore possible Service Level Agreement (SLA) negotiation strategies and implement these as part of the post-execution processes of the MIX protocol.