A data team designs a Databricks SQL pipeline with two use cases: (1) near-real-time ingestion of continuously arriving IoT sensor records from a streaming Delta source, and (2) periodic refresh of pre-aggregated sales summaries queried heavily by a BI dashboard. Which combination of Databricks table types is correct?
Answer & explanation
Correct answer: E
WHY E: Streaming Tables are built for continuously appending incremental data from streaming sources. Materialized Views pre-compute and cache aggregations with flexible refresh options (scheduled or TRIGGER ON UPDATE), which makes them ideal for heavily queried BI dashboards.
WHY NOT A: The assignment is reversed: Streaming Tables handle continuous ingestion, not Materialized Views.
WHY NOT B: Streaming Tables are designed for append-only ingestion, not for declarative pre-aggregated transformations intended for BI queries.
WHY NOT C: Materialized Views can be refreshed in near real time, but Streaming Tables are the architecturally correct choice for pipelines that ingest continuously arriving data.
WHY NOT D: Standard (dynamic) views recompute on every query with no caching, which adds latency and ingests nothing; Streaming Tables are insert-focused and not designed for aggregation summaries.