How dataset APPEND, REPLACE and ETL(Update Method only new rows) work internally? Does DOMO create a unique key internally and only update the modified or newly entered records from source?


Suppose, we have some trillions of data in SFMC(salesforce marketing cloud). First, I will create a app cloud connector to SFMC. Now, what is the best approach to have this working? Which will work faster way with no performance issue? If I schedule it to run on daily basis, as per my knowledge, APPEND will cause duplicates , REPLACE will completely truncate the dataset and reload it all over again, which will be time taking approach, and creating ETL(only new rows), will work for only dataflow, dataset still have all appended data? Please advise, for better performance how internally DOMO works, what would be the best option for million and trilions of data from SFMC.