Hello,
we are using SQL 2005 with SP1 and we have quite small cube (<1GB).
We have a couple of measures with aggregate function AverageOfChildren or LastNonEmpty.
End users have very poor performances while browsing cube using these measures.
When we replace agg. function with SUM, for example, performances become normal.
I saw simillar discusion on OLAP forum, but I did not understand what is the best recommendation for this situation?
Is this known issue, and how it can be avoided?
Best regards
Borko
Borko,
What's your aggregation strategy for this measure group? Queries to semi-additive measures are resolved by initially going down to granularity for the measure group's time dimension and subsequently computing the aggregate. Because of this, it's important to insure that you have an aggregation that includes the granularity attribute. If not already there, you can generally force the attribute to be included by setting the AggregationUsage property to "Full" in the cube editor. However, since this setting applies across all measure groups, you may want to set it, design aggregations for this measure group, and then rest the value.
I'm not positive that this will solve your issue, but this is where I would start.
-rob
No comments:
Post a Comment