Streaming and batch data transforms

What the exam expects

• Choose a streaming transform, batch transform or formula field from latency, sources and result shape.
• Configure a streaming transform: SQL with a primary key, a valid target DLO, joins, shared targets, delete-and-recreate.
• Pick the batch node (Append, Aggregate, Join or lookup, Update, Filter, Flatten JSON, Split) and valid input/output objects.
• Predict incremental versus full batch runs and their credit cost; design joins that avoid duplicates and skew.

Key terms:

• Streaming data transform: continuous SQL that reads one source record and writes rows to a different target.
• Batch data transform: a node canvas run on a schedule or on demand.
• Target DLO: the DLO a transform writes to, then mapped to the data model.
• Incremental run: a batch run over changed input rows only.
• Lookup: a batch join that keeps one row per left-side row.

Key facts

Check yourself