Automation of Cube Partitioning

When an analysis services measure group starts to have over about 25 million records, you should consider cube partitioning.  Partitioning allows you to split a very large fact table into smaller pieces.  This can certainly help with query performance and it can also help speed up cube processing.  Companies with very large data sets have found that they have needed to partition their cubes.

Why do we need automation?

Although Microsoft allows cubes to be partitioned, they did not implement any feature that allows the automatic creation of partitions.  An example would be a retailer that has a cube partitioned by month.  Each day, the cube has the dimensions updated and then the "current month" is processed.  (That way they don't have to process the entire cube everyday)

There is nothing built into Analysis Services that allows for the automatic creation of the partitions.  It is a manual process where the developer walks through a wizard and the partition is created.  The good news is that the automation of partition creation can be automated through the use of SSIS and the SSAS object model. (code)

What about automating cube processing?

Another issue is automating which partitions are processed each day.  There is nothing "out of the box" that allows a developer to dynamically process "the most current partition".  Even the SSIS Analysis Services Processing Task is a hard coded alternative.  The good news is that using code, the processing of certain partitions can also be handled.

 

Let us show you how to automate your partitions