Measures with Distinct Count Aggregation in AS2005/AS2008

Measures with Distinct Count aggregation are handled differently by the CubeSlice Relational ASSL and the CubeSlice OLAP ASSL. Distinct Count measures can be included directly in the local cube when using CubeSlice Relational ASSL. With CubeSlice OLAP ASSL, these measures must be converted into an equivalent calculated measure.

When possible, you should use the CubeSlice Relational ASSL when using Distinct Count measures, for the following reasons:

1. The calculated Distinct Count measure can slow down the performance of the CubeSlice OLAP ASSL local cube. This might not be noticed with smaller cubes, but can be quite pronounced with larger local cube files.

2. It is not always possible to create the calculated Distinct Count measure in the CubeSlice OLAP ASSL or it may be created with some browsing limitations. In order to create the Distinct Count calculation another measure has to be used. If there are no other measures in the same measure group a measure must be used from another measure group. If there are no measure groups that have all the dimensions as the measure group with the Distinct Count measure, the calculated Distinct Count will have browsing limitations. When browsed with the missing dimensions it will behave like any measure viewed with dimensions not included in its measure group, displaying the total value for all of its members.

If a cube doesn't have the required measure and level to convert the Distinct Count measure to a calculated measure, the Distinct Count measure and any calculated measures that depend on it are removed from the local cube.

The inclusion of the calculated form of the Distinct Count measure in a local cube can cause slower browsing speed for that local cube. All calculations take time, and if the local cube has a large amount of data, you will notice the extra time that is taken to perform the distinct count calculation. But a decline in performance does not occur in every situation. If you keep your local cube files to a reasonable size, you might not notice any degradation of performance at all. And there are many situations in which organizations need to use Distinct Count in order to provide significant information to their users.