I have one dataset that is connected to a data warehouse with a set of credentials that are default embedded as some type of global connection. Each report that is made using this dataset is often altered in desktop to use a different set of credentials. When it's altered, published to the cloud and made available, it often resets the credentials for other reports using that same dataset.
If I understand your question, that would currently be by design. The reason is that when you upload the dataset it says it is going to replace the dataset, which in my understanding means it is going to replace the credentials if they are different?
Did I answer your question? Mark my post as a solution!
Every report has a connection to some type of data source. Ours would be a database connection that is our data warehouse (Azure Data Warehouse to be specific). Within that database, you have various schemas broken down by client. Each client has their own set of credentials for near replicates of tables across each schema.
When we first defined a data source within a report, we created the original connection to said data source. This data source was saved as a global connection in Power BI. Whenever we reuse that connection, we change the credentials to the connection to match the client and the report/dashboard. When we make that change, it impacts other client reports using that same connection as if they are all connected to the same data source and set of credentials to said data source.
We want each credential to remain the same. Clearly, when you edit a report in Power BI Desktop, you have an option for editing the connection of the report only. Why in the world would that be changing all the other connections in all other reports when we publish the report back to the Power BI cloud?
We can't use a unversial user due to both security reasons and the fact Azure Data Warehouse is limited on how many queries can parallel query the database.