A Deployment Strategy Of Service Registries In Complex Network

UDDI service registry is an important part of SOA architecture. UDDI service registry provides the functions of service registration, service query and service management. Service provider submits services to UDDI service registry though service register interface. Service requesters search services which can meet functional needs through service search interface. Service requesters make loosely coupled interaction with service provider via UDDI service registry, the operation of update or change on service by service provider won't impact the original system.Traditional UDDI service registry is usually deployed in centralized structure. As the growing of service query and network scale. Traditional UDDI service registries incline to the performance bottleneck of the whole SOA architecture. If UDDI service registry failed, SOA architecture will be down and can't supply the service. Facing above problems, a hierarchical deploying strategy of UDDI service registry is proposed in large-scale distribution network; also, in order to resolve the conflicts among efficiency, reliability and load balance, complex network theory is introduced by the thesis. The thesis includes the following three parts:â‘ By comprehensive analysis of traditional UDDI service register, four-layer structure of UDDI service register, CSPS-UDDI model, is raised by the thesis in complex network. CSPS-UDDI model analyze the feature of each layer and make the partition of the whole network according to index of the shortest-path among nodes and community load balance exponent. Ordinary nodes complete to be UDDI service registry after community partition.â‘¡CSPS-UDDI service registry deployed model consider the influence when node join into network or quit network. An exchange mechanism between UDDI node and back-up node is proposed when UDDI node failed, which improve the reliability and robustness of the SOA architecture.â‘¢The complex network features of CSPS-UDDI model, including degree distribution, average path length and cluster coefficient, has been verified via simulation by MATLAB software. On the basis of features simulation, average service query time and community load balance exponent also has been imitated and compared between CSPS-UDDI model and other deployed models.