A Policy-Driven Distributed Framework for Monitoring Quality of Web Services

Quality of Service (QoS) characterizes the non-functional aspects of Web service, which is the key factor for satisfying service users and has to be carefully considered in service oriented architecture. There has been a lot of works on QoS description and service selection by QoS information. However, two issues have not been substantially explored in QoS management-how to collect QoS information from running services, and how to dynamically adapt service provisioning with run-time QoS information. Moreover, current monitoring approaches are focused on monitoring the service interaction in a single domain, while cross-domain monitoring has not been studied yet. In this paper, we present a policy-driven monitoring framework for collecting QoS information and adapting service provisioning for cross-domain service interaction. The monitor is built on our distributed QoS registry-Q-Peer. The monitor is user-centric, metric-oriented, feedback-enabled and loosely-coupled with service providers. Services are observed by capturing interaction messages, so the monitoring approach has no impact to service providers. The framework supports dynamic configuration of monitoring metrics for both single service and cross-domain composite service. It can be used as a reliable third-party QoS monitor for Web services.