Failed-tuple triggered blocking strategy for managing near real-time spatial data replication

Federal, state and local government agencies develop, exchange and maintain geospatial information for the purpose of land-related legislation and administration. Geospatial information change continually and geospatial datasets become outdated and hence unsuitable for decision support because of inadequate quality of data, and are costly to maintain. There is a growing demand for accurate and consistent geospatial information in critical sectors such as emergency response. Examples of such geospatial information include the address locations, cadastre (land titles/lots), Environmental Planning Instruments (EPI), and roads and administrative boundaries. Often geospatial data are not synchronized within an agency or across agencies, leading to inaccurate data in repositories and infrequent updates (if at all). Vendor software, such as ArcGIS provides a basic replication solution. However such solutions are vendor environment dependant. Furthermore, real world spatial enterprises have many complexities that lead to replication interruption leaving replicated databases in spatially inconsistent states. In this paper we use a near real-time event driven asynchronous replication mechanism for selective heterogeneous geospatial data based on a mediation framework. Within this framework, we have introduced a failed-tuple blocking strategy that overcomes some of the limitations faced in the real-time operation of the framework. The proposed method was applied to synchronizing cadastral geospatial information across multiple repositories, including a repository that stores data spatio-temporally. The repositories are used in near real-time Internet and other mapping applications. The result of this application is briefly discussed.