Are you saying that the Dataflow works fine EXCEPT for when it updates the target table?
If that is the case then I would suspect that:
1. You don't have a primary key defined in the target table. You would see a lot of other issues with this also.
2. There are no indexes on the target table that support the WHERE clause used by the UPDATE.
With regard to your billion row count in the monitor log, you can set the source table cache value to "No" and it probably won't cache every row from that table. It will however execute a SELECT statement for every row coming in from the flat file. In this case that might be a better option, but only if the master table has the right indexes to support the SELECT statement's WHERE clause.