Security on dataflow access/use

[po]

Hi,

Have power bi premium and can see access to dafalows currently disabled.

Woudl like to turn on access to this but only to selecteed users, such that only certain users can create and use dataflows?

How can this be best achieved?

Also what is best practice regarding dataflows.

For example say we have a calendar databae tale which woudl be used across multiple dataets in multiples workspace currently is it possible to have a dataflow workspace and other datasets in different workspaces just access that dataflow workspace so that import the object once via dataflow and all datasets access this dataflow verion of calendar?

Thanks

[jeffshieldsdev]

Unless you check enable there and a box appears prompting for a list of users to allow, enabling will allow anyone to create dataflows in Workspaces they have Contributor permission or greater to.

dataflows are secured at the workspace level, there's no other security. You'll have to think about your strategy a bit, maybe have a workspace called "Master Data Dataflows" for your date table and the like, give a few users Edit access to create and maintain and anyone that should be able to use those dataflows in their own dataflows or in Power BI Desktop Viewer permission.

EDIT: https://www.nickyvv.com/2019/11/quick-tip-sharing-power-bi-dataflows.html

[po]

Thanks for reply.

If have a dataflow which say pulls in calendar info, I'm assuming if people have datasets which wish to connect to this dataflow, fact dataflow will be executing only once say every time calendar updated (yearly or adhoc if need make manual chnage), that datasets referencing a dataflow don't cause it to be executed.

Is this understandng correct?

Thanks

T

[jeffshieldsdev]

Correct.  When a dataflow refreshes, it saves the output in ADLS behind the scenes, in CSV format...when consumed, Power BI Service or Desktop reads from this cache.

[po]

Thanks for the info.

We will alos have some fairly chinky datasets, tens possibly hundreds of millions of rows we would intend to add to via incremental refresh on dataflow.

Currently we have millions of rows being added to dataset via incermenatl refresh.

Fact csv files just wondering efficiency of large datasets with dataflows as opposed to incermental refresh into dataset.  Is there a recommendation in this regard as to which is better with huge volumes?

[jeffshieldsdev]

Dataflows are not a data warehouse, they're an ETL tool.  Remember dataflow output is stored in CSV files behind-the-scenes, so anything consuming from it will have to read-in that whole CSV file and process it with the Power Query engine--there's no querying as if a database, or query folding from the client.

There is the Enhanced Compute engine, which caches the data from CSV into a SQL store (again, behind the scenes--you have no visbility or control over how this works) but the SQL store will only be used by other dataflows in Premium Capacity when those sources are used as Linked Tables in the dataflow.  Power BI Desktop will always use the CSV-store.

EDIT: https://ssbipolar.com/2018/11/22/positioning-power-bi-dataflows/

[po]

Thanks agains for info.

We have recently got power bi premium - had pro before.

If we create a premium workspace with dataflow to say pull in objects can existing non-premium workspaces use the dataflow in the premium workspace?

Thanks

[jeffshieldsdev]

Non-Premium Workspaces can consume from dataflows in Premium Workspaces, however certain features (like Linked and Computed Entities) are only available in dataflows in Premium backed Workspaces.

I use non-Premium Workspaces unless I need those Premium features, to conserve consumption on my capacity.

https://docs.microsoft.com/en-us/power-bi/transform-model/dataflows/dataflows-premium-features

[po]

Thanks for info