Refreshing a dataset through a Power Automate flow from a PowerApp without providing excess access

[TKRA]

Hi,

I am currently in the process of developing a power BI report which should be accessible to a wide array of people, and I would therefore like to avoid giving too much access.

The report consists of a PowerApp which takes an input number, that is then written into a database. The input is then used to query a sizable amount of data which is used in the PowerBI report.

However, the issue lies in the fact that the dataset must be refreshed before the data can be visualized in the PowerBI report. I currently have two solutions to this issue:

1) When the PowerApp button collects the data for the PowerBI report, it also refreshes the report

2) After the PowerApp has collected the data for the PowerBI report, a separate button must be pressed in PowerBI to refresh the data

When i use solution 1), the end user must have at least "Contributor"-access in the workspace to refresh the dataset through power automate. However, when using solution 2), it is not necessary to give the end user more than read access in the report, Viewer access in the workspace and Read,Build access in the dataset.

While solution 2) works, it would be ideal to have the report be as user-friendly as possible by using solution 1), while also avoiding giving the end user too much access.

Below is the two same refresh a dataset-functions, but triggered from separate sources. Both were triggered by people with only read/build access in the report and viewer access in the workspace

When triggered from PowerApps, it fails with an "unauthorized" error

However, when triggered from a PowerBI button, it works

Please let me know if you have any solution to this problem, or know why there is a difference between these two triggers.

Thank you!

[lbendlin]

Two things to consider

- is there a gateway involved in your scenarios? If not, could it be (ie abuse it to refresh cloud data sources that wouldn't technically need a gateway)

- have you considered using Direct Query instead of Import Mode?  Or the new hybrid partition option ?

In either case the lack of native data write-back capabilities in Power BI will continue to make this an awkward user experience. Please consider voting for the ideas that have been raised about this.

[TKRA]

Thanks for the response!

Our main problem is that our full dataset is very large (apx. 3bn rows), so querying the data directly through PowerBI makes this proces extremely slow. It is admittedly a bit of a "hacked" solution, and having the end user press an extra button is not the end of the world, but i do find it very weird that the level of workspace access that is required to refresh a dataset, is directly tied to where the flow was triggered from.