Measures and calculations in cubes without natural hierarchies
In this part of my attribute relations series I will create a simple cube with only attribute relations to the dimension key and no user or natural hierarchies. After the cube is created I will look how cube measures and calculated measures will be affected.
One obvious reason for creating attribute relations in natural hierarchies is that they are the foundation for creating aggregations and better performance. There is also the point of not having wrong values in your cube fact measures but this is mainly caused by having wrong attribute relations in natural hierarchies and this will decide the distribution of measures values and not the totals.
I will create two simple dimension tables (Product and Time) and a fact table with a few records. After that we will create a cube without natural hierarchies and attribute relations in these hierarchies. We will only rely on the attribute relations between the dimension key and the attributes.
Run this script on a suitable database in management studio.
Tags: design