Keith,
Thank you very much for your input.

I realize that GUIDs are not 100% unique (there is only finite number of permutations (it might be a gazillion but still finite)), but for practical purposes it is pretty uniqe. Nothing is impossible, but some things are only possible in theory <s>

Do you know of any doc with benchmarks/statistics that show that joins are slower with GUIDs and that index inserts are slower with GUIDs. I know that what you say makes perfect sense, but .... :)

It looks like I've got a lot of thinking to do tonight. It would be nice if someone that loves GUIDs would pipe in because I am getting crushed here :) lol

Again thanks for your input.

Einar



>Also, joins are slower with GUIDs. Index inserts are slower. Indexes and data take up more space which means more page IO for queries.
>
>I'm not a big fan of GUIDs. The only thing that they are really useful for is merge replication.
>
>FYI: GUIDs are not guaranteed to be unique any more. A few years ago, MS changed the way that GUIDs are generated. It is still highly improbable that you would get a duplicate GUID, but it is no longer impossible.
>