Best Practice on extracting full population and succeeding data changes out of Oracle HCM

Summary:

Hi, we are revisiting how we are extracting Core HR data (specifically Personal Data and Assignment Data) out of Oracle HCM both full population and succeeding changes (80-100 data fields). From my experience, I think for high volume of data (e.g. 600k active employee), a file-based approach (HCM Extract) is still the most efficient way to do it. Is this still true or this can efficiently be done via REST API call?

Or it could also be a combination of HCM Extract for Baseline + REST API (ATOM FEED + Workers end point) for Near Real-time changes?

Let us know if there are guidelines on this or if you have experience how it is done in other customer.