Skip to main content
cancel
Showing results for 
Search instead for 
Did you mean: 

Score big with last-minute savings on the final tickets to FabCon Vienna. Secure your discount

Dataflow-based Datasets in service

The dataflows product is potentially very useful - I especially like the idea of being able to append historical data to incrementally refreshed data. However, one major drawback is the fact that in order to use a dataflow in a dataset, you have to download it to the desktop, and then upload it back to the service once you've completed your data model. It would be great if you only had to use the preview version of the dataflow in desktop, or even better, if you could do all modelling, DAX etc. completely within the service. If there are workarounds for this I'd love to hear them.
Status: Needs Votes
Comments
seary
New Member
This is a very frustrating limitation of Power BI. Please enhance the server version of Power BI so users can create reports that periodically refresh from Data Flows. As the author of this ticket implied, it is very frustrating to have to: use the Desktop version of Power BI to "download" Data Flow data into a DataSet; refresh the DataSet and/or Report; and then republish that report back to the server in order to get reports that are already on the server to "refresh". In fact, this is so inconvenient that I would possibly suggest to my superiors that we are better off (in some cases) using alternate products when periodic refreshing of data is desired. It seems to me that if periodic refreshing of data is needed on a server, that SQL Server Reporting Services (SSRS) or a third party charting solution might be more useful than Power BI (in it's current state). I know there are Streaming DataSets to try to overcome this, but I'm concerned that Streaming DataSets "may" have difficulty persisting data in some cases. I am also aware that you can use MS Flow (now called Power Automate) to create "actions" that generate or push to server side datasets on the fly, but that's a bit of a pain in my opinion.
mechairengineer
New Member
This is very much required as when the size is large, PBI desktop does not help
bing_bai
New Member
And can we refresh dataflow by automate?
fbcideas_migusr
New Member
Also, an external guest user can create a dataflow but there is no way to create anything useful (i.e. pbix) from that dataflow. How is a consulting partner supposed to create reports from the dataflows they just created on the client's tenant? Interim solution is to teach client to create a pbix based on the dataflow, publish to the service, from where I can download the pbix, edit, and use the Get Data in the service to upload the modified dataset/report. Rather a kludgy method.
greg18
New Member
It's definitely counter-intuitive to have the ability to create a dataflow in my Power BI Service workspace, but then not be able to use that dataflow as a source for datasets in that same workspace. I hope this functionality gets added soon.
daxissues
New Member
Completely agree. Very much needed
Eijaz_Sheikh
New Member
This is the difficulty i am currently facing. It means creating a dataflow in PBI Pro Service, connecting and downloading it to Power BI Desktop, modelling it and then copying the Source Code from Advanced Editor for each Entity and pasting it in same Entity in PBI Service. Wish there was a way to allow modification and update to a PBI Service Dataflow once connected via Power BI Desktop.
PowerBI193
New Member
Completely agree! very much needed.
Sujay_Dutta
New Member
Need the ability to build PowerBI dashboard from multiple SharePoint Lists and also API calls without having to use PowerBI Desktop. A lot of us are on the Mac and cannot use the Windows only PBI desktop! Please make the PowerBI feature rich as the desktop. Everything is API driven these days, not sure why PowerBI Desktop even exists!!
salvatore_cagl3
New Member
I agree with all comments above. It is a mess that it's not possible to create a dataset online, which is based on a Dataflow. There are a lot of people out here, that have large amount of data. This feature should be implemented very soon.