SNIP
>
>What I was attempting above was to determine if the size of the CDX was slowing the query down. I wanted to have one index only against the same data that all the other tests were run against.
>
SNIP
>
>Actually the Deleted setting can make a dramatic difference. I was doing a series of COUNTs against a large table and the times were 12 seconds, 15 seconds 9 seconds and just the other day they were averaging about .3 seconds. Once I figured out that SET DELETED was ON. I turned it off and the times returned to about .3 seconds.

Hi Dan,

I agree that the setting of DELETED can make a huge difference. BUT... in this case you said you had a single TAG and it was *NOT* on DELETED().
Yet you got a 3 second improvement running with DELETED OFF vs ON.
It is that that I find puzzling. I don't see why there should have been any difference at all!

I also agree that 25 seconds is way too long for a basic web form. But, if what you are offering is valuable to the user and you give them an adequate explanation/warning, then 25 seconds (or even 120 seconds, depending on) can be just fine.

I do think, for what you've described so far, that Walter's suggestion, or variations thereof, is likely the best possible.

good luck
JimN