Does anyone know exactly what circumstances cause Type 'N' rows to be stored in SYSCOLDIST over type 'F' rows. I can read the documentation of course and see it says Non-padded rows, which would seem to apply to VARCHAR columns since Type 'F' normally pad VARCHAR values out to maximum length of the VARCHAR. I don't recall seeing any Type 'N' rows so am guessing there are special circumstances. Not as simple as having a VARCHAR in the Frequent Value.
Query on a major subsytem to find Type 'N' rows found nothing. Used in multi-column (COLGROUP) freq value Stats with VARCHAR? Or some other situation? Very long VARCHAR (which normally would not be indexed) but could be collected by manually specifying?
DB2 Application Performance Specialist
CPT Global Ltd
Michael Hannan[Organization Members] @ Dec 13, 2018 - 07:30 AM (Europe/Berlin)
Michael Hannan[Organization Members] @ Dec 13, 2018 - 07:35 AM (Europe/Berlin)