The extended split index to efficiently store and retrieve spatial data with standard databases.
Jörg Roth · 2009
Geometric and geographic data have special demands on the database query mechanism. To store and retrieve huge amounts of geo data, special spatial databases thus offer geometric column types and spatial indexes. Even though spatial databases are getting more and more available, spatial operations are not standardized, thus applications cannot easily switch between different databases. Moreover, spatial databases are not available for all platforms. Mobile device platforms, e.g., usually only support standard databases without any spatial extensions. Our approach is thus based on relational standard databases and we introduce a spatial add-on that translates geometric queries to standard SQL. It provides a new spatial index, the Extended Split Index, which is optimized for the add-on. It especially avoids any index reorganization, makes use of one-dimensional non-spatial indexes available in SQL databases and heavily reduces the number of candidates that have to undergo further geometric checks. We demonstrate the strength of our approach with a performance evaluation based on more than 200 000 geo objects.