Last month I posted an article on the miss by Microsoft by not including a run-time surrogate key generator in the SSIS tool suite.
I received a couple comments asking why do I not use the identity column in the SQL Server database to be the de-facto surrogate key. This is certainly a good plan of action for using the SQL Server database, but the ETL jobs I am creating require the use of an Oracle database. Thus the identity column idea is not a workable solution for Oracle.
In my realm I have created an ETL methodology that is designed to be as database agnostic as possible however there will always be a need to customize the approach to the specific situation and request of the client.