>Check your xBase code carefully. It would never take under a minute on that hardware >configuration. Probably you're updating much less than you think.
Not with 10 million records, that's for sure. I started with 1 mio and put that up until my notebook began serious swapping data to disk. So what I was measuring was done mostly in memory. I don't see why SQL would gain a better performance factor with even more records. Maybe because of the raising spread of data for the month January 2005 as dDate is filled randomly.

>PS: I wonder how could you run new SQL construct under VFP6SP5.
I see this version number all over this thread, also with your postings. It has nothing to do with me.

Bye, Olaf.