We have an on-premise data source (SQL Server 2008 - I know its old...) housing our ERP data (Microsoft Dynamics AX 2009). We are currently using PowerBi to connect to this data using an import connection which is suitable for those current datasets/reports. I am now looking at building out standardised data sets for various finance related functions using Direct Query connections.
I am trying to find the best way to set these data sets up. My options so far are:
1. Simply connect to the server and data base using the PowerBi prompt and apply transformations/joins using the PowerBi native functions (steps etc.). Pretty much click everything together.
2. Write a TSQL query that performs all filters, joins and transformations and connect using that.
3. Pretty much step 2 but using Views.
I am not using SSAS as our SQL Server version is so old and I hear PowerBi has issues connecting to this (multi-dimensional connections are not supported until SQL Server 2016).
Basically I would like to understand how PowerBi handles Direct Query connection requests (i.e. user interaction with a report).
1. If the data set is clicked together through PowerBi would it effectively construct SQL queries based on the filters etc applied by the report and user and pull only the appropriate data?
2. If this is the case will it perform the same if I script the source data through using TSQL or connecting to a view - or is it irrelevant?
3. Or, does PowerBi always attempt to pull down the entire data set (table/s) from our database and then apply its steps (filtering/transformations etc.) after pulling the data down?
Am trying to ensure as much performance as possible. Direct Query is required as, for example, we will have people post journals/sales orders etc. and want to see the immediate statement listing.
If anyone has any input or links for me to read through to understand how PowerBi issues or constructs its queries that would be great.