PowerBI dataflow questions

[DebbieE]

I have been asked a few questions today about power BI dataflows and why you should use them.

namely how does it impact the data lake already set up ?

1. The Dataflow data is created in a 'black box' data lake. Can I check the file format on this. Im thinking its csv. But surely this should switch up to PARQUET which is where we are all heading?

2 If the above is where the data is stored when you create a dataflow. Where is it stored if you connect to the data in your Power BI desktop file? Is that just stored in VertiPaq  the  in-memory columnar database?

3. If you do set up a dataflow. does the data sit in the black box data lake and files AND also in the vertipaq column data store? Meaning you have it twice?

4 there is already a data lake holding PARQUET Files. How easy is it to not go with the 'black box' approach and actually use the Data Lake thats already been set up? 

Personally, I think the person Im speaking to will go against the use of dataflows if they arent in PARQUET format.

Also, If we end up with Data in the data Lake. Data in another Data Lake (Dataflows)  and Data in the Power BI data Store (3 storage areas) I don't think they will be happy with this. 

[aj1973]

Hi @DebbieE 

Creating a Dataflow is basically adding Entities(Tables) from one or multiple sources to it. The storage location of the Entities is ADLS(Azure Data Lake Storage). Once the Dataflow is created you will need Power BI desktop to connect it to it and to build the semantic model. From the semantic you can create one or more reports from Power BI desktop as well as from Power BI service when the semantic model is already published.

[DebbieE]

thanks for that but its the specific points above  I need information on to be able to talk this one through