Re: insert has high cost

Just playing a hunch here: Are there indices on the relevent tables on fields that have "clusters" of values? When this is the case, inserting a new index entry can be VERY expensive, as large portions of the btree must be rebuilt. As a general rule, fields that contain "Y/N" (or M/F, or similar) values should not be indexed. If you must, set the fill factor very low for these indices (30 - 40%), so the chances that an insert would require updates that cascade down the btree are reduced. Since fill factor is used only on _creation_ of an index, change the parameter, re-start the engine, and then drop and recreate the relevant indices.