>I've spend the last 4 hours looking for an answer on UT, I didn't think it was there. So at the risk of being wrong (and thus double posting) here goes:
>
>I'm trying to improve general performance of our application, but I've found something I don't understand. Following is just one example of slow queries, I've also seen on a tabel containing over 3,000,000 rows, this particular one contains 800,000.
>
>This is the Select I'm doing:
>
>Select *;
>from indibepk;
>where indibepk.clientnr+indibepk.indicnr==gcAlgClientNr+gcAlgIndicnr
>
>This yields a result of something between 0 to 15 rows (tops). Yet it's takes an amazing 6 seconds to create. Rushmore optimization is full (where condition is equal to index). There's a index on Deleted.
>
>If I Set Deleted to On, Rushmore uses the IX_DEL (Index on deleted) and takes 6 seconds. If I set deleted to Off, Rushmore is still fully optimized, but no longer uses IX_DEL, this takes 0,6 seconds to build the list.
>
>I understand the usage of IX_DEL, but why is the second run (set deleted off) 10 times faster?
>
>Again it appears to happen in more than one query and I feel I should be able to boost performance (if not dramatically).
>
>Any thoughts?

Geert,

What if you drop the index on deleted() ?

see http://fox.wikis.com/wc.dll?Wiki~VFPQueryPerformance~VFP and http://fox.wikis.com/wc.dll?Wiki~BigSystemPerformanceAndVisualFoxPro~VFP