I'm working on normalizing a database, and finding that I'm doing a lot of second normal form stuff, currently there are lots of tables that have what should be a record in another table flattened within itself.
I guess one would normally have (for example) a Vehicle_ID field that pointed to an auto-incrementing ID in the Vehicles table.
I'm pretty well convinced that the auto-increment ID won't work for us, because we will have disconnected applications, so I've been using guids. Is there anything to watch out for? I'm especially concerned about the best way to handle unique constraints, clustered indexes and how JOIN statements might be affected.
Any thoughts appreciated,
Mike
