It is a dimension table doesn't associate with fact table directly, but other dimension tables. We may have different scenarios to make a outrigger dimension.
Below is some example
- Reflecting reality relationship:
We may have a "home store" foreign key in the customer dimension to represent the store she first registered.
- Normalization:
To normalize a dimension, we may snowflake it. For this case, I think we would better call them snowflake dimension.
- Split a fat dimension vertically:
If there are too many fields in a dimension, it may be harmful to the query performance. We may consider splitting some infrequently used fields vertically. For example, we may sometime separate the Customer contacts from Customer dimension.
Outrigger dimension is acceptable in dimensional modeling but we normally avoid it if possible. It is because the performance concern. Outrigger dimension may be harmful to the querying performance.
No comments:
Post a Comment