Output Node: Write Recipe Results to a Dataset or External System

Note For an Output node, avoid using data that has multiple unique keys. This can occur when the output results come from an augment, where the each source dataset has a unique id field with the schema. Instead of an augment, ensure the output results come from joins between the source datasets.

1. In the recipe, select the Add Node button () at the end of the recipe.
2. In the Add Node dialog box, select Output.
   
3. In Write To, select the output type, Dataset, Existing Dataset, or CSV.
   • For Dataset, add:
     • The label, API name, and dataset’s app.
     • Optionally, if using sharing inheritance, select a Sharing Source to apply sharing rules.
     • Optionally, add a security predicate to restrict access to records with row-level security.
       

       Note After creating a dataset, you must edit it to make changes to its security settings; changes to security settings in the dataflow (rowLevelSharingSource or rowLevelSecurityFilter) or recipe (Security Predicate) have no effect.
     
   • For Existing Dataset:
     • Select the operation, Append, Upsert, or Delete.

     • Select the existing dataset to run the operation on. The append operation adds all data from the input to the existing dataset, and records can be duplicated. The upsert operation appends data from the input to the existing dataset if new and overwrites existing rows when they match. The delete operation removes input data from the the existing dataset when existing rows match.

       The upsert and delete operations only work with datasets created by external uploads and data syncs, which preserve the required isUniqueKey information.

     • Enter the dataset label and API name.
     • Turn on the disjointed schema option to ensure the dataset updates if the snapshot data schema changes. If the option isn’t selected, the recipe fails when the schema changes
     

     When using this option, remember that:

     • The dataset row limits apply. Consider adding and scheduling a recipe to delete old snapshot rows to avoid hitting the limits.
     • When the snapshot schema changes, and the disjointed schema option is active, the recipe takes longer to run as it reregisters all the rows.
   • For CSV:
     • Select the user who has permission to download the generated files.
     • Optionally, select whether to include a header in the CSV file.
     • Optionally, determine how to split the CSV over multiple files.
     
   • For Output Connection, see the relevant output connector help.
   

   Note The Dataset API Name can contain only underscores and alphanumeric characters. It must also be unique, begin with a letter, and not include spaces, end with an underscore, or contain two consecutive underscores. If another dataset in the same app uses the API name, the dataset is overwritten. If a dataset in another app uses the name, choose a different, unique name. If an API name isn’t specified, a unique API name is generated.
4. Click Apply.
5. Save the recipe.

To create the dataset, run the recipe. As you create more datasets, remember that your org has a row limit for all datasets in your org.