@Vidas: its totally clear to me that Da

@Vidas: its totally clear to me that DataIDs must not change the main problem that I see is that two Processing-Options work different and, from my point of view, not consistent if there is a built-in check for DuplicateKeys, shouldn't it work in all scenarios and not just in some? ProcessUpdate also has some inconsistent behavior in the 3rd example we add a row with invalid attribute relationship the SubGroup gets moved to the MainGroup of the new row, the MainGroup of the old rows does not seem to be checked (other wise we would/should get and duplicate key error?) in the last example, we remove that invalid row and the SubGroup gets moved back to its original MainGroup which indicates that the old rows are checked(?) @charlieMCTS: you are right, in a real-life scenario you would have seperate tables with a n:1 relationship for Product, SubGroup and MainGroup so, from a relational view, it would not even be possible that a SubGroup belongs to more than one MainGroup but this was just an example, in real-life you may also have other dimensions that are not commonly snowflaked where this problem could still arise