Wes <wespvp(at)syntegra(dot)com> writes:
> This appears to be very inefficient. B is almost two orders of magnitude
> larger than A. C is about 3-4 times as big as B (record count). My
> statement (with the same single 'B' table as above) produces:
If it's only a factor of 3-4 then the merge join should be faster. If it's
really two orders of magnitude (100x?) then the nested loop below would be
faster. I think in 8.1 (and I think in 8.0 too) the planner is capable of
coming up with both plans for the NOT IN query though.
--
greg