Implement RLS on a shared dataset

[pradeep467]

Hi,

I am trying to design a solution for below requirement. I have a dataset A which has 10 tables+1 security table for RLS. This is published into workspace A and all user access the report through App A. Now if I have to share this dataset with few users who would like to work on this dataset but apply their own RLS, below is 1 option

1. Share the dataset PBIX file with them so they can customise their own RLS and publish their own reports but that would duplicate the same set of core tables in PBI service.

2. I tried separating the 10 tables as a separate dataset and published to service but even then I am not able to configure RLS from dataflow and enable security

Please suggest if there is anyting that came up recently that can serve my requirement.

[pradeep467]

Thanks for the response Tom, yes your understanding of the issue is right, the audience that I want to share are different  and they also have need for different visuals, so instead of duplicating the dataset in 2 different workspaces I am trying to maintain a single source of truth.

I tried publishing a dataset with just main tables and withour any rls and then when I tried connecting to that dataset in service and tried adding a dataflow , PBI converted by dataset as direct query to AS connection and allowed me to join my stand alone security dataflow but security filter isnt visible.

[TomMartens]

Hey @pradeep467 ,

you to configure RLS for all audiences in the base dataset. It's not possible to apply RLS to a model that has been "converted" from remote to local, what happens if you connect to a Power BI dataset and change connection type from live to DirectQuery.

Regards,

Tom

[TomMartens]

Hey @pradeep467 ,

first, I have to admit that I do not fully understand what you mean by "... apply their own RLS."

Maybe these people want to share the dataset with their own audience(s) with different RLS applied by audiences.

If my understanding is correct, then the answer is simple. A dataset has a single set of RLS rules. Either it's possible to incorporate the required RLS rules into a single dataset (this is road I would take) or you have to create multiple copies of the same dataset.

Hopefully, this helps to tackle your challenge.

Regards,

Tom