Forum Discussion
"In Power BI PPU, Dataflow Gen 1 does not support direct query.
Hi lavendar ,
Thank you for your response, and I appreciate KNP for offering clear answers.
Performance issues with Gen1 dataflows, even with ECE enabled, are uncommon and often mean some query steps aren't folding as expected. It's a good idea to use the Performance Analyzer in Power BI Desktop or check the dataflow refresh logs to see if Enhanced Compute Engine is being used effectively.
For slow "Close & Apply" times in Import mode, this is usually due to merge operations that aren't folding to the source, so processing happens locally. To improve performance, try moving merge logic into the dataflow or into a SQL view upstream. Gen1 doesn’t require explicit storage since it uses managed storage internally.
Your architecture, which avoids direct connections to the production SQL DB, is good. The slowness is likely related to specific transformation steps rather than the platform itself.
Hope this helps. Please reach out for further assistance.
Thank you.
Super UserThanks for the additional info lavendar.
Unless the data is extensive, import is still an option. You can set up import mode with incremental refresh to ensure semantic model refresh times remain efficient.
Also, a direct query is possible with Dataflow Gen 1. You need to go into the Source step of the PowerPlatform.Dataflows(null) step to change the option.
Direct query should be avoided whenever possible. No matter what the source is.
Hi KNP Anonymous ,
Thanks for the solution, however i have toggled on ECE , dataflow is refreshed, still seeing very slow visula load times in Direct Query.If I try import mode , the power query close and apply takes too long , merge operations cause severe latency.
- Is this level of slowness typical for Gen 1 data flows even after enabling ECE?
- Why is import mode performing poorly even after applying paramters in power query , why is the close and apply load unresponsive/not loading?
- Is an explicit storage required for Gen1 to perform well?
- Our whole idea was not to connect directly to mssql prod db, as it can impact performance.
- KNP
Hi lavendar,
Is this level of slowness typical for Gen 1 data flows even after enabling ECE?
No, I've used Dataflow Gen 1 for dozens of solutions, some with very large fact tables, performance is not an issue.
Why is import mode performing poorly even after applying paramters in power query , why is the close and apply load unresponsive/not loading?
We would need to see the queries etc. to be able to diagnose that.
Is an explicit storage required for Gen1 to perform well?
No.
Our whole idea was not to connect directly to mssql prod db, as it can impact performance.
And this is a good decision. Your planned architecture is sound. Something else is causing your performance issues.
Can you do the merge operations further upstream, either in the Dataflows or a view in SQL?