Gimme ten minutes... I'll be right back...

One more question... Is it possible to add a single column to the table?

Ok... this takes 2 seconds longer (1,000,000 rows in 23 seconds instead of 21)... had to work around the "zero domain" on the LOG10 function to get this to work...

There would certainly appear to be a fly in the ointment  ... thanks for the catch and sorry for the mistake folks... I'll see...

I'm thinking it just might be something else because the following works just fine and without any error...

select datediff( hh, 'May 17 2002 9:53', 'May 18 2002 9:53')

Don't worry, Gail... we considered the source

...but when doing that I must maintain the sequence of un-duplicated numbers

Commando...

I think the quote above explains it...

A good ol' batch file that uses OSQL or BCP would be easy enough... and, then, the user could provide whatever filename they wanted as a parameter.

Choose wisely and carefully... you get one clustered index per table.  And, just because you have a clustered index on a table does not guarantee it will be used... you...

If the naming differences are that bad, generate scripts for all the views from EnterPrise Manger, drop all the views, recreate the views using the script.

This should do it... don't worry about the couple of strings you see... this is still pretty darned fast and they make it independent of the @@DATEFIRST setting... (you can...

Sorry... had posted a solution here (and deleted it) for hours that didn't take into account days that were on the weekends... I'll be right back with the correction....

This will give you whole days... modify the "ends" of the range to do hours... also, not affected by @@DATEFIRST settings...  very fast... very simple...

 SELECT (DATEDIFF(dd,@StartDate,@EndDate)+1)

       -(DATEDIFF(wk,@StartDate,@EndDate)*2)

      ...

Also, have you looked into using sp_AddLinkedServer... it, in fact, supports links to DB2.  See Books Online,

You could use Sp_ExecuteSQL but then you'd still end up with a table variable that cannot be made to use statistics...

If it were me, I'd use a temp table instead...