when to use OLAP cube

[Anonymous]

I have very high-level question about OLAP cubes relative to Power BI.  In what scenario(s) is it helpful to have an OLAP cube between a data warehouse (that is dimensionally modeled) and Power BI?

[v-luwang-msft]

Hi @Anonymous ,

The answer depends on your specific needs. OLAP’s functionality derives from how it analyzes data through multiple dimensions and processes large data sets in a way that is impossible using a standard spreadsheet. Using cubes, you could even make data discoveries that would not have been possible otherwise.

In comparison to two-dimensional relational tables, the cube shape represents the physical storage of multidimensional measures in an analytic workplace. This shape allows for rotation – a key advantage of dimensional modeling. Different analysts will need to view data in different ways, and OLAP allows them to manipulate and view the data according to their unique requirements. It makes for a powerful online visualization system with faster response time. Users can, therefore, slice and dice cube data using various filters, measures, and dimensions. It is also easier to find outliers in the clusters with OLAP.

To learn more details ,refer:

Multidimensional Data Modeling – OLAP Cubes

How to connect Power BI Desktop to a SSAS OLAP cube using the HTTP Data Pump

OLAP Cube

 

Did I answer your question? Mark my post as a solution!

Best Regards

Lucien

[Anonymous · Accepted solution]

Hello @Anonymous ,

As a person worked a lot with OLAPS and Tabular models. There is not really many cases that it is beneficial.

The main reason to use it is usually the cost. If you run the calculations in local server instead of premium capacity it can save some money. But still I would recomend use premium cloud.

SSAS Tabular is more beneficial and faster. And it is easy to develop due similarity to Power BI. I would choose that instead of OLAP, if you want semantic layer between DW and report.

This gives you also some speed and savings.

[AlexisOlson]

Can you give a bit more context?

Personally, I'd probably prefer to do all the dimensional modeling within Power BI rather than having multiple layers of models.