I don't have any performance data, but more on the usability part, the
general practice is to create an IDENTITY column and have it auto-increment.
It makes your PK's small and indexing them is also easier. Also, relating
them with FK's is easier too (since you need not store GUID's in all your
tables). Quering for information based on PK becomes easier if you use
IDENTITY columns. Some application have the concept of "prefixing"
information along with the PK too (for example, an order number might be
ORD001, where 1 is the primary key and the ORD00 is prefixed before it).
These sort of operations are easier with IDENTITY columns.
Please reply to the whole group only!
"Anand" <org> wrote in message