Posted by Doug on 03/22/06 20:25

the first thing the engine will do is grab all the valid id1's from
tableb, adn stuff them into a "temp table."

Then, because Table A has an index on ID1, it makes sense to sort that
temp table so you can go through TableA efficiently.

A question. Do you have a clustered index on TableA? Try getting rid
of the clustered index, and just having a non-clustered index.

regards,
doug

[Back to original message]