Unwanted dataset created on thin reports

[DTaraboletti]

I'm trying to create a shared dataset for multiple "thin" reports.   Following advice offered in various forum entries and online videos, I have successfully separated the data layer from the visuals/reports... but I'm confused by the creation of separate dataset files for each report file.   

As outlined in the following clip (https://www.youtube.com/watch?v=RxJBaYPWTOU)  at ~6:39, the thin report should not create a separate dataset file.  Thus, I would expect my lineage to look like this:

However, instead, I'm seeing an intermediate "unwanted" Dataset in the lineage... like this:

Viewed in list mode, you can see the extra dataset files:

Because of the extra datasets, the Automated, scheduled refresh (of the "Shared Dataset File") will not automatically refresh the reports... instead, each report "unwanted" dataset file will have to be separately following the successful completion of the Shared dataset file refresh.

Any idea why these extra, unwanted datasets are showing up in my lineage?

I'll note that within PBI Desktop, the data panel shows that 95% the tables are DirectQuery but a couple parameters are noted as Import...thus overall I've got a "Storage Mode: Mixed"  Is that the reason for the extra dataset files?  These parameters are present in the Shared Dataset and unchanged within the thin report files.  If this is the cause, can they be made to be DirectQuery as well?

Thanks in advance for your assistance.

[DTaraboletti]

Based on the documented "Limitations" above... closing this thread.   Thanks to everyone for your help.

[hnguy71]

Hi @DTaraboletti ,

At the 4:20 minute mark within the video you referenced, in your report copy, you can just delete the base model and then connect to your published model which would then be a live connection to your shared dataset. Afterwards, you should be able to use global filters to restrict data based on your parameter(s)?

[DTaraboletti]

@hnguy71 Thank you for the response.  I did delete the queries as noted at the 4:20 mark...but the "unwanted" datasets were still created.

Subsequent to my post, I've stumbled across something that I'm now investigating further.... I've used the text box with +value option.  Apparently, the +value option requires that Q&A be turned on for the file.  And, it appears that the Q&A setting may require a local "copy" of the dataset so that it can do indexing.  I'm in the process of removing these "text box" instances from the visual file to see if that fixes the mixed storage.   I'll post an update here with my findings.  In the meantime, if anyone has experience with this, please post your thoughts.

Best regards.

[DTaraboletti]

@hnguy71   Thank you for your response.   I'm afraid I don't follow (understand) your reference to "live connecting the parameters. How does one do that?

[hnguy71]

Hi @DTaraboletti ,

You mixed mode is likely the culprit of your extra models. Perhaps just live connect to them and filter at the report level?

[DTaraboletti]

@hnguy71   Further finding...  

I found, what I believe is the reason for the mixed storage.   In the report files, I am using a slicer control to dynamically set the Y-Axis of a chart as outlined in the following Guy In A Cube @GuyInACube  video

https://youtu.be/1eurc0EY2Xg?si=w4SZRfLINvu43iiZ

As noted in the video, to achieve this functionality, you must set a field parameter for the dynamic Axis. 

When the data layer is separated from the report file, the field parameter is being set to "IMPORT" instead of Direct Query mode.    So... progressing this troubleshooting, I'm now trying to change this to a direct query, but the Storage Mode property is grayed out.  

[DTaraboletti · Accepted solution]

As noted in this documentation...

Let report readers use field parameters to change visuals (preview) - Power BI | Microsoft Learn

The highlighted section (below) indicates that when field parameters are used, they are stored in local model/dataset.