Our OLAP Application Setup
Cube Partitions: Monthly
Average Number of Records per Partition: 23M
Active Partitions: 8
Distinct Count: Yes
Virtual Cubes: 1
Changing Dimension: 2
Operation: Incremental Update of All Partitions Everyday
Incremental Process of Cubes Every Day
BI Accelerator: Yes
Our main concern is that the knowledge base articles instructs us to re-aggregate the entire cube in order for us to work around the problem. The problem is we incrementally update our changing Dimensions everyday. To add to this, we measured how long it takes to full re-aggregate one partition – it took more than 45 minutes to finish one. Totalling all the time it takes for our process to finish, the time involved in the suggested work around would be impractical for us.
Is there another way to resolve these problems?
If we remove all user designed aggregations from all our cubes, will we still end up with the same problems?
I hope you give us some insights and suggestions with this problem.