I'd stay away from table valued functions unless performance is completely not a concern.

The cross apply should only return a single value such that you'd get more than one row,...

Hmmm. Not an easy one. I looked through the code, I generated the query. Everything looks ok, assuming the data types in the tables are as presented, with one possible...

Sounds like a problem inside the code of the stored procedure. Can you post it?

Assuming a straight SELECT query, at no point should there be a conversion from anything to...

Oh, OK. Well it sounds similar to a problem we have

SELECT ...

FROM p

CROSS APPLY (SELECT TOP 1 ...

...

Wouldn't the EndDate be a "+" 1 to make it a day larger than the EffectiveDate?

Also, I think that you may want to use the DATEADD() function because MS is...

SELECT TOP 1 *

FROM X

ORDER BY Something (ASC/DESC)

The ASC (ascending) will get you the first record, the DESC (descending) will get you the last record. Just remember that TOP processes...

If you've got an existing system, monitor it to see what, if anything, is behaving badly. Start from there. Once you've identified the issues, you can start to make decisions...

I agree with what everyone has written and, since I didn't see it, I'll add one more thing. The reason you have to be so picky about the clustered index...

You can't do that, but there is a filter tool in the Object Explorer and in Object Explorer Details window that will allow you to look at subsets of the...

Stupid question, I know, but is it possible that the SET NOCOUNT ON is, you know, off inside these queries? Depending on how you make the call, I've seen rows...

OK. I take it all back. I've done some more research in preparation for the tests I was going to run. The Ninja is right. It's a myth.

This is from...

One key difference, ignoring the "myth" for the moment, is that the constraint can be used as part of a foriegn key if you choose while the index can't.

Now,...

What John said.

You know, I've never tried, but I have to assume that it would involve a larger amount of data. Based on the information, I doubt you'd see a difference with...

No, I can't prove it. It's information I've inherited. I thought it was from an article by Kalen Delaney, but I can't find it. Here's a reference from Kimberly Tripp:...