In designing data loading for a Type 1 slowly changing dimension, which wording best describes the required action?

In designing data loading for a Type 1 slowly changing dimension, which wording best describes the required action?

• A. Delete existing rows and insert duplicates to preserve history.
• B. Update rows when the non-key attributes have changed and insert new records when the natural key is a new value in the table.
• C. Always insert new rows for any change.
• D. Never update existing rows; instead, raise an error on change.

Answer: (B)

Type 1 loading updates reflect the current state of a record without keeping any history. When data arrives for an existing business key, you replace the non-key attribute values in the existing row so the table always shows the latest information for that key. If the incoming data contains a business key that isn’t yet in the table, you insert a new row for that key. This combination—update for changes to attributes, insert when a new natural key appears—captures the Type 1 behavior. The other approaches don’t fit Type 1. Deleting and reinserting to preserve history would destroy simplicity and isn’t how Type 1 handles data. Always inserting for every change would create duplicates and fail to reflect a single, current view. Never updating and raising an error disables the dynamic refresh the dimension is meant to provide.

Type 1 loading updates reflect the current state of a record without keeping any history. When data arrives for an existing business key, you replace the non-key attribute values in the existing row so the table always shows the latest information for that key. If the incoming data contains a business key that isn’t yet in the table, you insert a new row for that key. This combination—update for changes to attributes, insert when a new natural key appears—captures the Type 1 behavior.

The other approaches don’t fit Type 1. Deleting and reinserting to preserve history would destroy simplicity and isn’t how Type 1 handles data. Always inserting for every change would create duplicates and fail to reflect a single, current view. Never updating and raising an error disables the dynamic refresh the dimension is meant to provide.