Power Bi - Inefficient Backend Development ?

[Alexrk]

Hello, I have this question:

Lets assume that a Power Bi dashboard developer has at his/her disposal a backend database developer to create any backend solutions they might need.

If that was the case, you will never have to face situations where you need to create lenghty DAX scripts and join too many different tables or views within the Power Bi dashboard to solve the challenges presented by the databases, because any challenges you face can actually be solved with the backend databases.

Would you agree with the above statement ?

[m3tr01d]

Hi @Alexrk 

I agree with @Daryl-Lynch-Bzy. He already said a lot of interesting things. I want to give you an example

Let say that you sell boats and you have access to all boats sold by the competition. You will be interested to look at what is your Market Share in the Industry.

Market share, being defined as a SUM over a SUM will be difficult to store it inside your database. You can store it but it will for a specific granularity and set in stone. Market share is a non-additive KPI in this case.

What if you want to look at the market share in all these contexts?
- By Product?
- By Year? Last Year, Last 3 months, Last 4 weeks
- By Segment of the industry?

Using Power BI, you'll definie only Once the KPI as a Measure and you'll be able to build different visuals and easily change the context by Drag and drop fields from your dimensions.

You build one Dataset with a lot of KPIs with predefined business rules that you document then you share the dataset to others where they can build different reports always using the same business rules and logic.

Hope it helps!

[Daryl-Lynch-Bzy]

Hi @Alexrk - yes and no.  Data should be transformed "as far upstream as possible and as far downstream as necessary" (Roche's Maxim).  But the key here is "transformations".  This should not include "Data Analysis".  Power BI using DAX provides a better experience for Data Analysis.  To understand this you need to see the SQL Queries that are dynamically created by the DAX engine to extract data from Star Schema.  By simple drag and drop, slicer selections, Power BI uses DAX to generate thousands of lines of SQL.  So when someone say an SQL developer can solve the business requirement, I laugh.  Too many times, I have seen SQL Developer who does not understand DAX hard-coding TOPN, PIVOT, UNPIVOT, CASE WHEN AGGREGATIONS and JOINS when SSAS Tabular, Power BI and DAX would have more flexibility.