How to restrict authors from a subset of data

Question & Answer
, , ,
1

In our business case, we want to give the access of Analysis to our customers. So that they could create and publish their own dashboards. This part we have already figured out. The next part is to give the authors department-wise access.

Our parquet files are stored in a S3 location which is queried through the Athena tables. Quicksight will be connected with the athena table as data set. Say the table has three columns as following
| id. | department | Name

|. 1. | Sales. | A
|. 2. | Marketing. | B
| 3. | Developer. | C
| 4. | Sales. | D
| 5. | Sales. | E

Now we want any author from a department should only get the access to the same department on the Analysis page to create his/her dashboard. Say if Author A is opening the Analysis page he/she should get row 1,4 and 5.
Initially we thought that the row-level security security could be our answer. But from the documentation we found that row level security is more for the readers directly on the published dashboard. Can anyone please help how exactly we can incorporate same row level security kind of use case for our authors also?

2

RLS is applied to the dataset, thus it should apply to authors working on analyses based on the RLS secured dataset as well as readers using the dashboard.

Notice the word SHOULD. We have never tried that at my company. Give it a try with a very simple dataset that gives users just the count of records (and not the raw data itself). Let us know.

1 Like
3

Hi, @KaushikR. Did @cgreenacre’s solution work for you? I am marking their reply as, “Solution,” but let us know if this is not resolved. Thanks for posting your questions on the QuickSight Community Q&A Forum!

4

Hi @Peter @cgreenacre : Thanks for your kind suggestion. The solution actually worked for me. When I created the dataset from actual datasource(Athena for us), I had used the filter to segment them and create separate datasets on QuickSight from a single table. And saved them. When we invoked respective datasets from an Analyses, the filters were working as RLS

1 Like
5