Great explanation, you have a talent of explaining something very complex in a very easy to digest manner!

Thanks so much Matthew. I'm new to leveraging dataflows and this was a great primer for their use vs a dataset. Thanks again!

Thanks Matthew, very well explained!!!

Thanks for this BI Polar, very good video ! 2 questions: 1) Can a dataflow be consumed in multiple workspaces? 2) Is there a suggested way to organise the dataflows? For example would you create one dataflow per database and include all its tables or would you create a separate dataflow per entity?

Wonderful Matthew, thank you!

Finished watching

Great intro and explanation. Thx

Great Stuff Matthew. Like the new channel. Look forward to future videos.

Much better. I like your problem-solution approach. Thank you.

Excellent video! I'm wondering though. 1. What's the difference between a Power BI Desktop Dataset (tabular model) and Power BI Service Dataset (seems basically identical, but w/o modeling capability). 2. You cannot create relationships between entities (tables) in Dataflows correct, instead it's more used as a data source to connect to to then build relationship within Power BI Desktop?

how relevant is this now? surely i can create a dataset and that can be stored on the PBI service, and users can connect to the dataset from power bi desktop, im not seeing the big value add here.

Hi Matthew, I couldn't find anything on this: The primary keys defined in my dataflows don't exist anymore when I import that table in power BI desktop. Is that unsupported? Should I need to recreate the primary keys in the dataset by Table.ReplaceKeys? (I need primary keys for performance on merging and aggregating tables in the dataset)

Great explanation, Matthew! I've got one question, am I able to share just a couple of dataflows within a workspace with a specific user or is it "all or nothing" kind of sharing?

So.. the example you've provided uses an web api as a source, publicly accessible via the internet. What about using an on-prem database or file system as a source? Will it embed credentials in the data flow as well?

I have one Dataset , I published to the service into its own dataset workspace , I have a second work space for PBIX files (workspaces are segregated, I can apply different security measures across the two workspaces, i.e. no user can see the dataset workspace but only has read access to the dataset, same user can see the workspace or App where the reports reside) ; multiple PBIX files use this one Dataset (one central location, refresh , security) , has all the benefits of a dataflow plus the vertipaq engine for performance ; lean report files referencing one dataset published in the service (live connection to the dataset in PBI Service) ; is this a better approach or would you still recommend dataflow in this scenario ; only downside? Initial dev is done locally vs in the cloud , limited to you local machines resources

Thank you Matthew, it's very clear! Can we insert columns and measures in dataflow like we can do in BI desktop?

Can I define relationships inside a DataFlow? Or should that be inside a DataSet only?

Very nice explanation. Thank you

What edge does data flow has over published Power BI dataset. What I mean is I define a Power BI dataset in PBI desktop without VIZ and publish it. Other Power BI reports will use this Power BI dataset as its data source.

Nice and useful

Awesome Matthew, this is great explanation. With DataSets, you mentioned that behind the scene is Analysis Services Tabular and hence I can modify the semantic via its exposed XMLA endpoints (I had a requirement to do so with a project). With Dataflows, could that model be changed in a similar fashion (it sounded like it is no longer using Analysis Services behind the scene).

Hi Matthew! I appreciate your helpful video. I got a problem. I'm dealing with the huge dataset (around 25 million data rows) and increasing fast day by day. I only have a Power BI Pro account. I build a dataset in power bi desktop, then publish it into power bi service. But it takes a long time to refresh in pwbi services. I tried to use Incremental Refresh but it does not work in my case. So I wonder when I used dataflow, it can help me to reduce refresh-time in pwb services? Thanks for reading this comment and best wishes!

Hi Thanks for the video. I got a question. If we prepare our datamodel, do all the transformations, etc. and then share this prepared model as a Power BI dataset which other people can use, and modify it when necessary we have a single point of truth with modifications. So where is the use case of dataflows in this case. I do not really get that. Could you please elaborate? Thanks so much!

Hi BI Polar, Thank you for this, is there any data on the Pro v Premium licences differences and limitations and finally - If you've built Power BI reports directly from source and wish to change current reports to a Dataflow and also Dataset usage is this do able without having to re-build the reports?

Hi, is it possible to use dax tables created in power bi desktop as a datasource for data flows ?

Hi Matthew! About cloud sources for Dataflows, like Azure SQL database for example. When I tried to do the connection, i am obligated to install and configure the gateway to make it work (even though it is a cloud source). As far as you know, are we obligated to use the gateway with cloud sources like azure sql database? Because If we don't use dataflows, we can schedule refresh without the gateway, so it is much easier to maintain. Hope you can give us some guide. Thank you for your videos!

Doesn't sound like Power BI is easy for business folks to use. Must go to Wikipedia to search for jargon every other sentence. Tableau seems much more intuitive.

I was really able to understand the fundamentals of what a dataflow is compared to a dataset: you explanation was very clear and helpful! Thanks :-)