Creating Your First Multidimensional Cube - Getting Started with SQL Server 2012 Cube Development - page 69

Note
Defining the correct attribute relationship is always considered as best
practice. With the attribute relationship, Analysis Services can use less
memory when processing. Accessing the attribute is much more effective
and this leads to better query performance. Aggregates are also more ef-
fectively handled when attribute relationships are defined.
Defining the right type is also fundamental. Rigid relationships should be
used whenever the relationship between the attributes does not change.
In our example, a quarter never changes to year and a month never
moves from one quarter to another.
If you would try to process the dimension in order to be able to browse it at this point,
then you would get an error stating that a duplicate attribute key exists in the dimen-
sion. The reason for this is that the there is no unique value for the Month and Cal-
endar Quarter attributes. January exists in all years in Q1 and Q1 exists in every
year. What you need to do to solve this is to change the attribute KeyColumns prop-
erty to use a composite key. To do this, follow these steps:
1. In the dimension editor, click on the Calendar Quarter attribute.
2. Then in the Properties window, click on the KeyColumns property and the
... button.
3. In the Key Columns window, add the CalendarYear column to Key
Columns .