At each employer I now start by writing a data extractor that takes one client's data and loads it into my local DB. Usually includes some 'skip' or filter opti

At each employer I now start by writing a data extractor that takes one client's data and loads it into my local DB. Usually includes some 'skip' or filter options to make the process faster. And some transformations like anonymizing PII. With PG you can bypass the FK checks locally by temporarily changing the replication mode from origin to replica. Though if you don't eventually copy the referenced records you may run into FK errors during normal operations.