Thank you for replies. The transform is a simple one, all work on the data is done using updates after the update.  We are currently importing a table that takes 2h 40m and need evidence to support our theory that being provided with a csv will speed this up. The  CSV would be created by those responsible for the Oracle database that we are importing from and they are disputing that an extract, move and load would be quicker than Oracle to SQL DTS.  I was hoping that there were official figures to back up my theory to add to the sample testing that I can do myself.