>>>Solution #2 uses a different starting number for each database and sets the increment at two. Say DB1 has a seed of 1 and DB2 a seed of 2. When each of them use the increment value no numbers will ever be repeated and you should be safe for future growth too. #2 seems more elegant IMO.
>
>What happens when you want to introduce another site (read "database") into the replication scheme?
>
>I've not been following this thread so I don't know all the details. What about introducing a site id and making the PK a composite of the site id and the IDENTITY?
>
>-Mike

Or, how about adding a uniqueidentifier to each table and using that as the replication key?