The pipeline review uses Aster Supply's orders and Blue Heron Rides' events. Both sources contain records that need explicit handling. A graph can show dependencies and counts, but you still need to read what each node promises to publish.
Reuse streaming table, materialized view and expectation rules from this unit. A dropped record and a failed update are different operational events. The first changes the accepted output; the second prevents the update. Decide which one the stated requirement allows. Also distinguish a stored result from a query definition. A materialized summary needs refresh behavior, while a regular view reads its current sources. None of the questions asks you to infer production throughput from this browser, because the local engine does not model distributed scheduling or cloud storage latency.