Search operations in nonspatial databases can be optimized using SPATIAL indexes. This is true for spatial databases as well. With the help of a great variety of multi-dimensional indexing methods that have previously been designed, it is possible to optimize spatial searches. The most typical of these are:

MySQL uses R-Trees with quadratic splitting for SPATIAL indexes on spatial columns. A SPATIAL index is built using the MBR of a geometry. For most geometries, the MBR is a minimum rectangle that surrounds the geometries. For a horizontal or a vertical linestring, the MBR is a rectangle degenerated into the linestring. For a point, the MBR is a rectangle degenerated into the point.

It is also possible to create normal indexes on spatial columns. In a non-SPATIAL index, you must declare a prefix for any spatial column except for POINT columns.

MyISAM supports both SPATIAL and non-SPATIAL indexes. Other storage engines support non-SPATIAL indexes, as described in Section 12.1.8, “CREATE INDEX Syntax”.