I fully expect this to be driven by target technology platform being used - so let's presume we're talking Oracle database...
I'm interested in what your view points are regarding the use of WHERE NOT EXISTS to drive the insert of new records into hubs, links and satellites vs NOT IN (or another approach). Presume the staging tables don't only just contain the unique new data, but contains modified and in some staging tables it contains unchanged data too (due to full extracts provided that we have no control over in some cases). Is WHERE NOT EXISTS seen as generally acceptable for the inserts into the RDV tables in all cases?
I also expect the volume of data being loaded and the proportion of change vs what us loaded already to be additional factors.
As always, I'm interested in your views.