Using Aggregations to Performance Optimize a Cube - Getting Started with SQL Server 2012 Cube Development - page 230

on the Bikes level and or on the Year level can—together with relevant hierarch-
ies—improve performance considerably.
In the FirstCube project that is built on the AdventureWorksDW2012 database,
you can take the DimDate dimension as an example. The lowest level of the date
dimension is on the day level. The date dimension then contains the month, quarter,
and year attributes and a hierarchy defining their relationship.
The month level of the DimDate dimension is built up of all the dates belonging to
that month. The quarter level is built up by the months belonging to that quarter and
the year is built up by the quarters. This means that if you often perform analysis
on the month level, then you want to pre-aggregate your fact data at this level. An
aggregation on the month level will also help if you want to view data on the quarter
and year levels as well. Consider this table showing the number of unique members
on each level: