Report performance with Row level security

[ani_informa]

Hi 

I have a requirement of row level security in my Power BI report. I have fact table "Sales" with more than 10 million rows. There is a dimension table named "DimStore" and it contains unique row for each store (StoreKey) and there are 2K unique storekey rows in dimension table. My StoreUserBridge table contains 2000 * 5000 users = 10,000,000 rows.

I want every store user should be able to see data belongs to their store only and that's the intend behind using row level security. I just wanted to know whether fact table with more than 10 M rows, StoreUserBridge with 10,000,000 will impact the report performance or not ?

Is there any document, post available that can provide the benchmark, pros n cons of using row level security

[Anonymous]

I'm not sure if you have come across this article but it's worth mentioning. 

Making sure your model is optimised is obviously step number 1 as RLS applies filters in dax queries in the background

Row-level security (RLS) guidance in Power BI Desktop - Power BI | Microsoft Docs

[Anonymous]

Hi @ani_informa,

I do not so recommend you to create 'one: one' mapping RLS for the ultra-large amount of records, it obviously will affect the performance due to duplicate records.

I searched for power bi relationship bi but not found documents that mentioned how to mapping rows without affect performance.
Since power bi data model stored data at AS instance, maybe you can take a look at AS tabular model RLS related document:

Optimizing RLS performance with the Query Store 

In addition, you can also consider creating a weak relationship mapping between user and store, then you can add filters and conditions in your RLS expression to help to map records more accurately.

Relationships in analysis services tabular models  

Regards,

Xiaoxin Sheng