0

So i've finished my dimensional modeling, it resulted in 2 business process, 1 simple with only one fact table and a few dimension, the other one a bit more complex with 2 fact tables (related in a similar way has Invoice and InvoiceRecord) and a lot more dimensions.

My question now is how to start building the OLAP cube(s), one for each Business Process? Or one for each Business Process and for each fact table?

RASMiranda
  • 371
  • 6
  • 19

1 Answers1

1

You need to consider all the fact tables and dimensional tables for creating a common star schema. You should consider creating single cube unless fact and dimension pairs are not interrelated at all. It all depends on your design.

Prabhash Jha
  • 330
  • 2
  • 12
  • Both the star schema are done for each of the business process, also the ETL is done in SSIS. I'm now moving to SSAS, designing the cube(s). Here are the star schemas, i only have them in portuguese but i hope it helps. [Operational Bussiness Process Schema](http://i.imgur.com/Cn1oQp8.png) (Facts: Servicos & Entradas); [Finacial Business Process Schema](http://i.imgur.com/bzKXW67.png) (Fact: Resultados). So do you suggest creating only one cube? Based on a data view containing every table? I started playing with SSAS, and automatically created 2 data view, for each business process schema. – RASMiranda May 28 '13 at 09:04
  • 2
    For this case my suggestion would be to create 2 different cubes for Operational business and Financial business. Facts: Servicos & Entradas seems to have no link with Fact: Resultados and its related dimensions. Therefore if created only single cube, measures from both these facts would be irrelevant to most of not-linked dimension. I would suggest to create two different cubes unless there is some requirement to show results from both facts together. – Prabhash Jha May 28 '13 at 16:31