Forum Discussion
How do you prevent a dataflow from creating duplicates?
Thanks for the reply, @cchannon. Very interesting. Is that different from the default key that is created for Dataverse tables (e.g., Name)? Also, how do these map onto items being imported? If, say, the data is coming from an Excel spreadsheet, how would those rows be identified, or is that handled behind the scenes by the dataflow?
Make sure your target table has a Key set up that uniquely identifies each row (maybe just the name? or a composite of name and type?) Dataflow will pick up on the key on the table and for all the rows to be imported it will use the key to upsert instead of just inserting.
Thanks for the reply, @cchannon. Very interesting. Is that different from the default key that is created for Dataverse tables (e.g., Name)? Also, how do these map onto items being imported? If, say, the data is coming from an Excel spreadsheet, how would those rows be identified, or is that handled behind the scenes by the dataflow?
Yes, it is different from the default Primary Column because the default does not enforce uniqueness, whereas a configured Key does.
To set one up, just navigate to the table you want to configure in the make.powerapps UI and you'll find Keys in the left nav. Select that, then create a new key, and you can then select all the columns you want to use as a composite to form a unique key. For example, in the screenshot below, I have a table called Assignment and I am creating a key that enforces uniqueness by a combination of Name, Due, and Type:
As far as linking them together with your import, that's actually the easy part! Dataflow automatically detects that there is a Key associated to the table you are importing and takes the field mapping you provide to build that key and validate it against what is already in the table. If the key for the imported row already exists in the table, Dataverse does an Update, otherwise it Inserts (and you don't need to set up any of the logic - it just works!)
That's awesome! Unfortunately, I'm currently using a Dataverse for Teams table, which apparently doesn't allow users to define keys in tables. 😢
Great to know this is possible, though. Much appreciated.