Include the granularity attribute of the time dimension in aggregation designs of measure groups with semi-additive measures, according to query patterns
This rule analyzes aggregation designs in measure groups that contain semi-additive measures to verify that the granularity attribute of the time dimension has been included.
Best Practices Recommendations
You should include the granularity attribute of the time dimension in some of the aggregations, according to query patterns. Queries to semi-additive measures are always evaluated at the grain-level of the time dimension. For example, the calculation engine will automatically calculate a query for the inventory count in Q1 at the month level, if the grain of the time dimension is months, and it will then compute the result for Q1 "on-the-fly". Because of this query time behavior, aggregations above the month level will never be used for semi-additive measures. If additive measures are present, higher level aggregations will be used and therefore should be defined as well.