Currently, Dataiku offers the choice to either overwrite or append data during dataset updates, yet it lacks the capability for a user to perform an upsert on their data. An upsert operation, which merges the functions of updating and inserting, enables users to harmonize their existing dataset with new or modified data. Unlike the current practice of users opting for complete dataset overwrites an upsert could present a more efficient alternative.
Existing workarounds involve resorting to custom SQL code or utilizing a combination of an append recipe and a window recipe. However, these methods are not user-friendly for those seeking simpler solutions, often leading them to default to a complete overwrite.
It's worth noting that other users have also expressed the need for this feature in different posts, although it hasn't been specifically addressed here. [1] [2] [3] [4] [5]