In profiler I see reads and duration.

But CPU is still 0, is there any way to get a closer look at that?

Since the queries are not taxing the system, can...

Now that is very interesting.

Can you explain why that is in more detail?

Does it have to do with the order in which it accesses the data on the table for...

Makes good sense.

The main question was "Why does this one behave differently than this other one here", and that does help to explain that.

Thanks again!

A few more random things:

Looks like one of the servers is Enterprise edition and the other is Standard and the execution plans are completely different between the two.

So I am...

Yeah, I figured as much.

I didn't know if it would be a simple "Oh yeah, last month patch XYZ came out and it addressed sorting...".

I'll gather that up and post...

Thank you, I will look into Idera and grab their trial to take it out for a spin.

I'm very open to other suggestions as well as just general guidance.

Thanks!

Crazy business rules, we have to get the ENTIRE thing, or toss it all.

Also we can't allow users to come back and pickup where they left off as we don't...

Grant Fritchey (7/30/2009)

---

OK. So how are the bogus fields identified? In client code or TSQL? If on the client side... 20 calls might not be that expensive. If on the...

So where I'm sitting right now is thinking (from what was suggested above):

Pass that data to a sproc, the sproc builds a tally table (from that suggested article, thanks btw)...

Here is the whole process.

User enters a LOT of data in a survey, nearly 200 some questions.

Some of the responses are validated against some business logic, things like:

Age = 18...

Gotcha, thanks.

Anything WRONG with the trigger approach?

Granted it may slow things down if there are lots of updates, but is that the only real concern?

The slowdown due to extra work...

So what is I leave it blank and don't use that option?

What does the default to?

Also I don't think I fully understand that.

If I grant execute rights on a sproc...

Of course!

AH!

Thank you, that makes much more sense now.

For some reason I was thinking the END statement was sufficient to signal the SPROC was done.

So basically the only real option is to setup additional SQL Servers and program to spread the work out among them?

Good point... I can see that killing performance during massive data loads as well...

Thanks.