It is also asked in this: https://groups.google.com/d/topic/druid-user/xPdSmowsQD4/discussion
Is it possible to upsert an entry to a data source ? While ingesting data I may have duplicate entries and I don’t want them both to be in the data source. Also after some time an entry’s metrics could be change but all dimensions and the timestamp should be the same. An answer in the link says it possible but it’s expensive. How can I exactly handle these two situations ?