Did you had any other more realistic tests in mind??

Well the real pk of that table is priceless for that query :

Select * from dbo.MainTable M inner join dbo.Adresses A on M.UniqueId = A.UniqueId and A.RealPk = (Select...

I won't be slowing down for you honey . Did you...

BTW you should really more looking into this :

to test the code :

exec dbo.pr__SYS_MakeAllRecordProc 'TableName', 1, 0

refresh the list of procs and see go see tablename_insert ...

You'll have...

This is the missing part of Noeld's solution :

WHERE P.CORRESPONDENCE_ID=@corrid

What he meant was to pass multiple ids at once instead of calling this proc multiple times.

How come uniqueid is not unique??? Is it a fk of some sort?

This works on my pc... but I got many tables... so the end gets truncated... maybe you'd be better off doing loops of 20 views at the time. Also...

I think he got that point earlier .

You can always try to rebuild the data into the iso standard using string functions and retry the import.

That's why I added an offset parameter.

But this is always a valid option :

Select top 33 percent from dbo.MyTable order by Newid(), this will be a true random sampling...

I haven't always had a free hand, but when I proved everyone wrong around me (to the last argument), they gave it to me and never asked it back... best...

Sorry but if the management (assuming untrained management here) forces you to do stuff like that, you're probabely better off starting to find a new job, or we'll be seeing...

This is one way to do it... you can also use delete from join syntaxe. But it yields the same execution plan.

You want to be the best, don't do like the rest .

So what was the problem?