Re: group by clause filling up temp space

Look at the BYTES and CARD of the HASH JOIN. A quick look tells me that these numbers are way too huge! The number of bytes estimated for the hash join is 6,958,868,504,868 (roughly 6.5 terabytes!).

My first thought is that a NESTED_LOOP would be a better operation. But the optimizer is not choosing it, why? Are there indexes on a.sid and b.sid? If not, you might look at them to see if a better execution plan is the result. Received on Tue Aug 30 2005 - 14:56:54 CDT