Question 37

A fraud-detection team needs new transaction events ingested into a Unity Catalog table with **end-to-end latency of a few seconds**, processing only newly arrived records and recovering automatically from failures. A second reporting team consumes the same data but only needs an aggregate refreshed **once an hour** with always-accurate, fully-consistent batch results. You are building both flows with Lakeflow Spark Declarative Pipelines. Which combination of objects and pipeline mode should you choose to satisfy the fraud-detection requirement specifically? ```sql -- Fraud-detection ingestion (latency: seconds) CREATE OR REFRESH STREAMING TABLE transactions_bronze AS SELECT * FROM STREAM read_files( 'abfss://[email protected]/txn/', format => 'json' ); ```

• AA streaming table in a pipeline running in **continuous** mode.
• BA materialized view in a pipeline running in **triggered** mode.
• CA streaming table in a pipeline running in **triggered** mode scheduled every 4 hours.
• DA materialized view in a pipeline running in **continuous** mode with a 2-hour trigger interval.