Portable High-Performance Indexing for Vector Product Format Spatial Databases

Geo-spatial databases have an overall performance problem because of their complexity and large size. For this reason, many researchers seek new ways to improve the overall performance of geo-spatial databases. Typically, these research efforts are focused on complex indexing structures and query processing methods to capture the relationships between the individual features of fully-functional geo-spatial databases. Visualization applications, such as combat simulators and mission planning tools, suffer from the general performance problems associated with geo-spatial databases. This research focuses on building a high-performance geo-spatial database for visualization applications. The main approach is to simplify the complex data model and to index it with high-performance indexing structures. Complex features are reduced to simple primitives, then indexed using a combination of a disk-based array and B+-Trees. Test results show that there is a significant performance improvement gained by the new data model and indexing schema for low to medium zoom levels. For high zoom levels, there is a performance drop due to the indexing schema's overhead.