Different User Same Datasource in a pbix, is Dataflow a solution?

[Anonymous]

I need to connect to Salesforce with 2 different usernames in the same pbix. As we cant do that in Desktop since the credentails gets overwritten. 

In order to overcome that - 

1) can we publish/create the dataset in the service and then connect to it thru desktop 

2) create a dataflow in the service and then connect to it thru desktop

What is advisable in the scenario. 

thanks in advance.

[Anonymous]

Hi @Anonymous ,

You can connect the dataset on Service through Desktop but i doubt if it will solve your problem.

Please take a look at below artical about how to switch datasource through parameters.

https://www.red-gate.com/simple-talk/sql/bi/power-bi-introduction-working-with-parameters-in-power-b...

Best Regards,

Jay

[Anonymous · Accepted solution]

Thank you all for the help! 

@Anonymous - the link is helpful but it dint help much in the scenario. My data source is saleforce, only the URL can be passed as the parameter and not the credentials. 

What worked for me was - 

1) Created a dataflow in the service with different username in the connection.

2) Connected to dataflow and saleforce in the desktop(both with different credentails) and published in the workspace. 

3) In the service (as I learnt) - 

Each user can only have one set of credentials per data source, across all of the data sets they own... 

So I had to take help of a teammate to publish one data source and other was published by self so as to enable refresh to work. 

Now data is getting refreshed from the same source for differnt users.

[tctrout]

I am not familiar with Salesforce as a datasource, however I encountered a similiar issue with another source that may be helpful.

Our workaround was to create a service account that had access to all the information and then published the solution with this service account.  This way the service account's is used as the user.

[lbendlin]

Neither one is an option. Maybe if you create two separate dataflows then it might work.

You'll have to wait for the composite data model feature that was promised for later this year.