Instead of storing metadata across 18 tables, the PODS4 upload process saves it using DB2 columns of type XML and does an INSERTquery into four tables.
We discovered that there is a point where aggregate index costs—more-expensive INSERT and DELETE operations, for example—outweigh a new index's query speed-up benefit.
This query tests that every stored procedure is prefixed with "usp_" followed by "Insert", "Update", "Delete", or "Select", followed by at least one entity name.