cancel
Showing results for 
Search instead for 
Did you mean: 
OSAdmin
Community Manager
Community Manager

BLUEPRINT BULLETIN

ARCHITECT FACTORY

Release Date:  12/08/2020 Updated

Prepared by: Peter Fugere
Severity: Medium

Summary:   Performance is affected if a customer needs to have the same entity in multiple cubes because of restructuring or frequent changes.

Purpose:    As entities move between cubes, an entity needs to be created in the new cube, all data reloaded to the new entity, and then the entity in the old cube needs to be deleted. This is a lot of work.

Key Recommendations:

 

Option 1: Simple Approach   The first option is the simple approach which is to include the entity in multiple cubes. There is a significant performance impact of this option. Because there is only one cube, the default aggregation would have all data unit members potentially sub-total to the parent entities. As the data unit grows above 1 million records, the performance will continue to degrade, and there will be no reporting benefit. In this case, consolidation rules could be used to limit the detail consolidation. For example, at every entity above base level, product could be captured at product line, and the detail below that level remains only in base entities.

 

Option 2:   The second option is the paired cube solution. There would be an entity dimension with all base members. For each cube, there will be a second dimension that has its hierarchy but uses the base members of the first entity dimension as its base members. For each cube in the original design, create two cubes. All dimensions must be the same except for the entity dimension. One cube has the base entities dimension, the other cube has the hierarchy dimension. Then the cubes must be linked, by making the cube with all base entities the child of the other cube. This effectively creates a base cube with all entities for its parent. The users would only ever be in the parent cube, so they do not see all entities.  While this will not remove the need to copy the data, all data could be copied by a rule or data management job. The linking of cubes must then be done by rules.

Option 2 works very well for large applications but adds a level of complexity.

 

Benefits and Costs

Each option has a pro and con, but the performance must be weighed with the administration of moving the entities.

 

Version history
Last update:
‎08-24-2021 01:02 PM
Updated by:
Contributors