>I have a program that runs overnight and processes records from a Parent table, for each of these records it processes around 17 Child tables.
>
>The program uses Rushmore all the way through using SEEK's and SCAN WHILE's, and for each Child record found, it inserts the info into a Table (the program is effectively making a copy of the DB).
>
>This program runs 7 times, and for each time is processing different Parent records, in total it runs from 17:00 to 23:25.
>
>I took the 1st run which is 1112 records from the Parent table and obtained the time taken for this process. It is taking (on average) around 0.679 seconds per Parent record, I know this is quite fast, but I want to know if it can be speeded up!
>
>Does anyone have any suggestions on the best way of tackling this, or am already doing it the fastest way, I thought that using SEEK's was the best way of getting info quickly.
>
>Suggestions greatly appreciated.

To start, I would suggest as much memory as possible (for buffering data), turn off all unnecessary services like Virus scanners and "scheduling software", use separate Drives for "reading" and "writing" (preferably SCSI), defrag your disks at off hours, turn off any "mirroring", insure that the OS where the app is running is set for "optimize foreground processing", try a faster CPU ... after that, you can think about tweeking the program (the tables you are writng to should be "unindexed" at this point; indexing, if required, should be done after the fact).