It's something we've looked at but (since we're dealing with temporal data) the only way we know which are recently added and which were already there is to add some of batch id. However, since we're aiming to store a couple of billion rows, it seems a bit of a waste.
I'm not sure I understand what you're saying ...
Are you saying you don't have any sort of unique index on the base table? If you do have a unique index then I'm not sure why you can't join between the inserted table and base table.
Obviously (?) trying to access your billions of rows of data, without a unique index, could also be problematic re: performance and selectivity ... ???
