Slowly changing dimensions
Slowly changing dimensions (SCD)
Dimensions in data warehousing typically contain relatively static data about entities such as geographical locations, customers, or products. Given that modifications in dimensions happen much less frequently, in unpredictable manner and outside of the regular update schedule that applies to fact tables, these types of dimension tables are called slowly changing dimensions (SCD).
Type 0: History isn't preserved. Dimension attributes reflect original values.
Type 1: Dimension attributes reflect latest values (previous values are overwritten)
Type 2: Every version of dimension member represented with separate row in the table usually with columns that represent period of validity
Type 3: Keeping limited history for selected attribute(s) using additional columns in the same row
Type 4: Keeping history in the separate table while original dimension table keeps latest (current) dimension member versions
When you choose an SCD strategy, it is the responsibility of the ETL layer (Extract-Transform-Load) to keep dimension table(s) accurate. Look at SCD strategy setting in SSIS.
Comments
Post a Comment