Bounding-box-only R-tree indexes
Default R-tree indexes use leaf pages to store bounding box representations of the data object. This type of R-tree index is called a bounding-box-only R-tree index.
To set up your own DataBlade® module to support bounding-box-only indexes, you must implement the RtreeInfo function with the operation strat_func_substitutions. You might also need to redesign your strategy functions that occupy slots 5 and up, if you want them to behave differently at non-leaf pages. This is because you cannot distinguish between leaf and non-leaf items in a bounding-box-only index. For more information, see The RtreeInfo function.
- The R-tree index is smaller, saving both disk space and the time to build and maintain the index.
- Bottom-up build performance is improved because memory and temporary dbspace usage are reduced.
- The log space that is needed to update the index is reduced.
You might want to override this behavior if your table contains other large columns in addition to the column that is being indexed with the R-tree index. For more information about the BOUNDING_BOX_INDEX index parameter, see R-tree index parameters.
Functional R-tree indexes are not bounding-box-only indexes; they store the data objects themselves in leaf pages.