Showing results for 
Search instead for 
Did you mean: 

Modeling patterns

10 - ETL
10 - ETL

There are a couple of dimensional modeling patterns that you see out there, that I'm wondering if Sisense supports correctly.

1. Snowflaking - (I don't mean Snowflake the database) - having rollup, lower-grain dimensions hung off of other more detailed dimensions.  This is pretty common.  In our early Sisense experiences, this produced unpredictable results, but maybe it's a supported pattern now?

2. Multiple dimensions off the same fact foreign key.  This is not common, but occasionally you have a foreign key ID in your fact table, that is referenced by two similar dimensions.  Think of a date ID which points at two different date dimensions.  It's not a good modeling practice, but sometimes it happens.  You'd want both dimensions to be referencable by dashboards using the cube.  Supported?

3. Fact table being referenced as a dimension.  Let's say you have a very detailed large fact table, and also an aggregated fact table summarized at the customer level.  In dashboards, you want to use some of the metrics on the summary table as dimensions (row cells in a pivot), not as summable facts.  Think if it as dimension attributes that just happen to be numeric.  Does Sisense support this?