A Dynamic QoS Driven Pluggable GIS Grid Services Architecture

As a natural evolution of GIS (Geographic Information Systems) Web services GIS Grid services (GGS) are defined with common behaviors and semantics using a certain interface description language, which makes the Service-Oriented Architecture (SOA) and its runtime environment more flexible and manageable. A new concept of Pluggable Service is proposed in this paper, which aims to separate the static binding between services and resources when published and to realize the dynamic binding at runtime, resulting in the improvement of the service virtualization, together with better resource sharing and service performance. In this architecture, a kind of Selector is used for selecting the currently best resources in a dynamic GGS Virtual Organization (VO) according to the combination of time-dependent Qualities of Services (QoS) and the resource requirement of specific service providers. Based on the selection result, dynamic binding to the best-fitted resources will be established, making the service available to clients, without any differences from the traditional approaches. The prototype system is built on top of the Globus Toolkit 3.2 (GT3), the de facto standard middleware enabling Grid services. The effects and performance of the system, evaluated through simulation, exhibits the benefits obtained by this architecture.