Did you cast the eachid value to the correct data type? If there's mismatch and the split value as higer priority, the clustered index value will have to be...

Outside the sandbox is the real term here. It's nice, but overkill, but nice .

So that can't be made to work??

set statistics io on

set statistics time on

dbcc dropcleanbuffers

dbcc freeproccache

go

update1

go

dbcc dropcleanbuffers

dbcc freeproccachego

go

update2

go

set statistics io off

set statistics time off

Noeld never gave me hard times.., He's more helpfull than anything else. Sushila tends to be more chatty which is hardly giving me any hart times at all.

Later Bulldozer...

I tend to disagree on this one :

--I use this table for many other string operations as well

CREATE TABLE [Numbers] (

[PkNumber] [int] IDENTITY (1, 1) NOT NULL ,

CONSTRAINT [Pk_Number]...

That's a good start. Now if you run select statement in query analyser you should see them added at the end of the list in the profiler.

Make a request to steve... I'm sure he can arrange that. In the meantime we'll make a pleasure of calling you bulldozer

How do you go around that?

Strange... I tried dropping but the buffers and the proccache between each runs and the 2nd query always got the best speed (alternating the queries). what am I missing??

Yup it's the same... which ever statement I run second has the best speed (only dropping the buffers between runs, not the proccache).

Nice catch... if it works once it works twice . Is it any faster speed wise?

Start / all programs / ms sql server / profiler

Open a new trace on the server in question (don't start it just yet), then in QA run Select @@spid. Go...

Can this help you?

Create table #A (id1 int not null, id2 int not null, primary key clustered (id1, id2))

Insert into #A (id1, id2) values (1, 2)

Insert into #A (id1, id2)...

Try again. Start a tracem create the temp table then leave the table alone for 30 minutes. Then redo the query. If it fails go back to the...

HTH.