Question 40

You are modeling a `dim_customer` dimension in Unity Catalog from a CDC feed using Lakeflow Spark Declarative Pipelines `AUTO CDC ... INTO`. The business has these requirements: - Auditors and analysts must be able to reconstruct **what a customer's address and tier were at any point in time** (full historical record). - Each historical version must carry validity columns so the **current** row is identifiable (active record has a null end timestamp). - Out-of-order CDC events must be sequenced correctly by an event timestamp. Which **two** statements correctly describe the SCD design you should implement? (Choose TWO.)

• AUse SCD Type 2 (`STORED AS SCD TYPE 2`), which keeps all versions with `__START_AT`/`__END_AT` validity columns and marks the active row with a null end value.
• BUse SCD Type 1, because it preserves the complete history of every change to the customer record.
• CUse the `SEQUENCE BY` (or `_sequence_by`) clause so out-of-order CDC records are applied in the correct order.
• DUse SCD Type 1 with change data feed disabled, because Type 1 inherently stores every prior version.
• EUse SCD Type 3, which is the only type that can store the full unbounded history of all past values.