Direct Query Preview feature with multiple datasets

[BethC901]

I am a new PBI Pro user with Premium.  I created a workspace with Premium content and am begininng the process of modeling against an Azure SQL database. 

With the recent release of the direct query using multiple datasets, should I be exploring that space or should I put all the tables into 1 dataset?  In theory, I could have multiple datasets broken down as:

Membership

Revenue

Fundraising

Development

Constituents - this data would be access by each of the other datasets

All PBI reports would access multiple datasets

-or-

is it better to have all the data in a single dataset using Live Connect?

[BethC901]

Thank you for your reply.  It would make sense to have separate workspaces for security reasons.  But then where would I put the Constituent table (and a few others) that will be needed by each workspace?  

[collinq · Accepted solution]

Hi @BethC901 ,

Yo ucan have more than one workspace using the same dataset.  Then when you share each workspace/report/app do NOT share the underlying data and that will keep users from getting to data that you don't want them to see AND allow you to write reports off of one dataset.  

[collinq]

Hi @BethC901 ,

Here is a good article explaining some of the differences.  Power BI – Live connection vs. Import, comparison ... - Microsoft Power BI Community  To me, the real question to answer is "What am I doing to the data?"  and the more you are modifying or working with it the more you get away from Live or Direct Query.  

As for the dataset question, putting the data together is one thing but displaying it and whom you are displaying it to and what "security" you have around it might be the driver between what you combine or not.  It might make more sense to keep the outputs separate into different workspaces/apps (like membership, revenue, etc).  If everything is "wide open" to everybody then one big dataset might be easier for you.  You can still have security based on whether or not you let them reach the underlying data.