Summaries
Dimension hierarchies are also convenient for organizing data summarization. Normally this would mean that a parent’s value is an aggregation of the values of its children. By aggregation we do not necessarily mean summation of data but some way to derive the data from the dependent values. Since aggregations are defined within a dimension, this means that the cells values in the whole slice of the cube via the parent member have to be computed from the cell values in the subgrid corresponding to a restriction to its children. Computing every cell in this slice seems to be very expensive, not computing it affects query performance.
The optimal solution must lie somewhere in the middle.