Re: SELECT/JOIN performance on temporary tables depends on timing of index creation

2005-11-25 Thread Allan Miller
Gleb: Aha. OK, we tried using OPTIMIZE instead of ANALYZE, and that does indeed update the Cardinality of the index, the way you would expect. Thanks very much for figuring this out! I really appreciate the help. Thanks again! Allan "Gleb Paharenko" <[EMAIL PROTECTED]> wrote in message new

Re: SELECT/JOIN performance on temporary tables depends on timing of index creation

2005-11-24 Thread Gleb Paharenko
Hello. I've checked this and found that ANALYZE table really doesn't work, but OPTIMIZE table made its work. In case it won't help you send to list complete definitions of you tables and queries. Allan Miller wrote: > Hi Gleb, > > Thanks for the quick response. Unfortunately, ANALYZE

Re: SELECT/JOIN performance on temporary tables depends on timing of index creation

2005-11-23 Thread Allan Miller
Hi Gleb, Thanks for the quick response. Unfortunately, ANALYZE TABLE does not appear to affect the Cardinality field of a temporary table (it is still NULL, even with rows added). Only by creating the index itself after the insert seems to make the query faster. I also tried FORCE INDEX and it

Re: SELECT/JOIN performance on temporary tables depends on timing of index creation

2005-11-23 Thread Gleb Paharenko
Hello. > Why is it that adding an index before or after the addition of data >makes a difference on SELECT performance? Check whether the problem still exists if you run ANALYZE table after inserts in case you have already created indexes. FORCE_INDEX in the SELECT clause should help as