Showing results for 
Search instead for 
Did you mean: 

Share your thoughts on DirectQuery for Power BI datasets and Azure Analysis Services (preview)

Hit Reply and let us know what you think of the DirectQuery for Power BI datasets and Azure Analysis Services.  To learn more about this feature, please visit this blog post or our documentation.


Here are some areas that we'd like to hear about in particular:

  • Performance
  • Query editor experience--the remote model query doesn't show up in the query editor and only in the data source settings dialog. What are your thoughts?
  • Navigator experience
  • Thoughts around governance and permissions for models that leverage this feature
  • Nesting models, i.e. building a composite model on top of a composite model
  • Automatic page refresh for live connect in composite models

Thanks and we look forward to hearing your feedback!


- The Power BI Modeling Team

327 REPLIES 327

thanks for the feedback!

Regular Visitor

This is really a step towards real "self service BI", but with a big bottle neck... "data governance". With RLS not allowed in feature, data usage goverance is lost. People can create there own data-sets, and share with others without "owner" of data-set having no visibliity visibility left upon with whom data is shared. And, bigger treat is IF we have external B2B tenant option enabled, person can share content to external unauthorized users.  
Please consider enabling RLS (effective user within data-set connection) for this feature, it has a huge use case in Power BI "self  service" community.. and with RLS it will really become a jewel of feature. 

The RLS rules in the original source will be applied.

What if OpenAccess Power user A creates a composite model, saves in 'My Workspace' and shares with person B who has 'restricted access' to parent sales. Now, user B can see all data as RLS defined in parent model is not carried on to composite model created by user A. This is big governance beach ache and I am not able to solve it. 
Any advise.

RLS is not being applied _at the moment_ as we are still in preview. It will be applied soon. RLS rules on central model will be applied even for B

Advocate II
Advocate II

@jeroenterheerdt @tessahurr 

I wanted to let you know that I posted additional details pertaining to my team's errors here. 

@Zarek @frostystoo : I believe these are the same errors you are also seeing: 


Regular Visitor

Been waiting for this capability and looking forward to it ever since it was announced. My initial plan of use was to break apart some larger complex datasets into 'child' datasets that could then be combined back into the larger parent dataset but still allow other organizational reporting areas to leverage some of the specific content of the child datasets to prevent the need from adding a bunch of tables that would be unnecessary to the application. 


Initially this was looking good. However, I ran into a limitation with the model being directquery, namely the limitation of passing more than one million records to an external source.


My setup was two separate Power BI datasets with a one to many relationship (I also tried bidirectional just to see) between a fact table from each. The visual renders fine until a dimension from the second dataset is added to the first. If filters are applied to the second dataset so that the resulting record count is below 1,000,000 records everything is good. Even though the first dataset has far fewer than 1,000,000 records, it would seem the second is trying to pass 1,000,000+ across the relationship and thus erroring out due to the directquery limitation.


This is unfortunately something that would likely frequently be an issue for us when trying to build out a robust enterprise model leveraging datasets from different functional areas to more effectively and efficiently leverage our premium capacity.  


I think this can still be useful for some targeted uses but not as much from a large governed data model.


Unless I'm doing something wrong - then I'm all ears! 


Another issue I came across was that I can no longer get directquery dataflows to work whereas I could before. 

Hi @clementskr !
See my blog post - DirectQuery gives weired queries with enormous big DAX...
Go for "Import"!

“Composite” arrived – CAUTION! 



I share your sentiment sir!

Thanks for your reply. I pretty much always use import. However, if HR has a Power BI dataset that they have built with all of the enterprise semantic layers built into measures and likewise Finance has a Power BI dataset (both Imported on their own), and I want to tie the two together with a relationship between the two without rebuilding all of the already approved measures, which is what this feature allows me to do, I'm limited by the direct query max rows limitation of the relationship is of high cardinality. Both sources are imported in themselves but tying them together forces it to a direct query source. Obviously, a single imported dataset of a full enterprise could be made, but that seems like a maintenance and refresh nightmare. 


So as far as I can tell, connecting large functional specific datasets to other large functional specific datasets isn't really enabled (depending on what that relationship is and the cardinality of such) because of direct query limitations.

Ok, I was not aware that also there are limits when the direct is to an tabular in memory...
I also had some other issues to connect datasets.
Stacked datasets 

Helper III
Helper III

Hello again.


For some reason im getting this issue whenever I publish reports using the new Feature (Multiple PowerBI Dataset)


Can anyone help me whats going on?

I'm having the exactly same issue on my end. I believe it is related to problems with build permission. The publish work correctly when you connnect to datasets located in a workspace you are member of. The above exception occurs when you try to publish a report which utilizes Direct Query connection to a dataset you only have build permission to but are not a member of a workspace it resides in. 

I have a support ticket opened on this but it is progressing painfully slow.

not sure, does this happen with other "normal" datasets as well?

Helper I
Helper I

Hi @tessahurr!
You asked for stacked datasets, I tried 😉
Let's do not think about if this makes any sense for real purpose...

I did build up 3 seperate datasets and then joinded them with an imprted table:
1) works fine
2) Access permissions to each single dataset no problem
3) DAX query is fine
4) Of course you have some how to handle duplicated dimension tables... (Geo)


Then I made of this complete model a dataset again.

This shoud be the base of a user report.

But I got some access issue... "Single Sign-on"?


Multiple DatasetsMultiple Datasets


Single datasset build of multiple datasetsSingle datasset build of multiple datasets

Helper I
Helper I

Ok, got it.

Yes that would give a strange model - exactly what I try to point out in my BLOG, that design will get even more important, what to join, how to join, which data to use...

Helper I
Helper I


A great innovation long awaited.
But also a risk, as even on perfect designed models and small dataset it gives weired queries with 3.000 lines or even 50.000 lines of DAX!!!
Read this:


“Composite” arrived - CAUTION! 



Frequent Visitor

Hello. I have a use case for needing more than 3 chain lengths for direct query to AS. Our Operations department uses multiple applications in their day to day activities. We have a data warehouse where we have slowly connected most of these applications and modeled their data according to our processes for easy reporting. We still have some applications not yet added to our data warehouse and some data sources which will never be added to our DWH so we have utilized data flows to access these data sources. We have created about 4 certified datasets for the data from these applications connected to our DWH and shared these within our organization for employess to use if they desire. Now our Operations department wants a set of KPIs in a single report based on data from all of the various applications they use. Up to now, we have had to develop a single BIG dataset containing the same tables from all of the certified dataset plus the data from the above mentioned data flows. This is so we could create the single report with multiple bookmarks, buttons, and slicers. This BIG dataset has been a nightmare to maintain (over 100 measures, over 60 tables) and we would much rather compartmentalize the data into these smaller certified dataset which would be easier to maintain and manage changes. I attempted to create the same BIG model using the new DQ for AS feature but since it contained 4 datasets and about 15 dataflow tables, it would not publish to the service. Please @tessahurr increase the maximum number of chains!

We are in a similar situation. We have one large dataset in order to support cross-departmental reports. However, this leads to difficulties with refresh scenarios.


Preferably, we could have process specific data-marts that refresh individually and could be joined together in a composite model when needed. Better still would be if those marts could be fed by dataflows. However, in this scenario, we would have already used up most/all of the chain before the user ever gets to it.

thanks for sharing your use case, this is great input!


For those that might have missed it:

Update December 18th 2020: we have just released Power BI Desktop version 2.88.702.0 which fixes the two most common issues we have seen being reported:

  • As part of this feature, significant changes were made to the authentication flow for live connections to Analysis Services which applies even when you do not have this preview feature enabled in Power BI Desktop. Users were not able to recover from entering incorrect credentials or selecting the incorrect credential type and would see error messages such as 'Unable to connect'. You can now either select 'Retry' or use the Data Source Settings dialog to update your credentials.
  • Converting from a model that is live connected to a DirectQuery model failed if the model contained a measure-only table and the following error message was displayed: 'The given key was not present in the dictionary'.

Helpful resources

MBAS on Demand

2021 Release Wave 2 Plan

Power Platform release plan for the 2021 release wave 2 describes all new features releasing from October 2021 through March 2022.

July 2021 Update 768x460.png

Check it out!

Click here to read more about the July 2021 Updates

Power Query PA Forum 768x460.png

Check it out!

Did you know that you can visit the Power Query Forum in Power BI and now Power Apps

Top Solution Authors
Top Kudoed Authors