Configuration settings for Power BI Desktop

Configuration settings for Power BI Desktop

Adjust the following Power BI desktop configurations to enhance DirectQuery efficiency, resolving slowdowns originating within the Power BI environment. You need to ensure that the machine is performant and has enough CPU and memory to install the Power BI desktop.

Maximum Number of Active Connections Per Source (DirectQuery)

  • This setting configures the number of parallel queries sent to the DirectQuery data source.

  • It controls the degree of parallelism when directly querying external data sources.

Steps to configure

Power BI offers a setting to limit concurrent queries to control the query load on the data source with DirectQuery. The "Maximum connections per data source" configuration lets you define the maximum number of queries DirectQuery can send simultaneously to maintain system performance.

By default, the Power BI Desktop initiates a maximum of 10 concurrent connections per DirectQuery source.

To modify this default, you can adjust the setting in Power BI Desktop via File > Options and Settings > Options > Current File > DirectQuery.  

Increasing the "Maximum connections per data source" can improve report responsiveness in scenarios such as:

  • Reports with a high number of visuals on a single page.  

  • Reports accessed by a large number of concurrent users.  

Note:

Increasing the connection limit results in a more significant number of queries being sent to the data source. The recommended configuration depends on which dashboard is loaded in Power BI Desktop. If a Power BI dashboard or Power BI worksheet generates more queries, then setting a higher number is recommended.

For example, if running a dashboard with 10 widgets and the Power BI Dashboard generates 15 queries, then setting the maximum connections per data source to 15 or 16 is recommended. This allows all the queries to reach Kyvos, which can handle the scale effectively.

image-20250416-100048.png

Parallel Loading of Tables

  • This setting dictates the maximum number of concurrent tasks the Power BI engine executes during data refresh, optimizing post-transformation performance.

  • Power BI automatically determines an optimal value, up to the defined maximum.

  • Adjusting this setting influences the degree of parallel table loading, affecting semantic model refresh speed.

Note:

This setting interacts with the "maximum number of simultaneous evaluations" setting.

Steps to configure:

To adjust the "Maximum number of concurrent jobs" setting within Power BI Desktop, follow these steps: File > Options and settings > Current File> Data Load >Parallel loading of tables

  1. Choose from the following options:

    • Default (6): This option allows Power BI to utilize up to 6 concurrent jobs for data loading.

    • One (1): This option turns off parallel loading, restricting data loading to a single job, which can significantly slow down refresh processes.

    • Custom (1-30): This option enables you to specify a custom number of concurrent jobs.

    • Recommended: 15 concurrent jobs.

image-20250416-100144.png

Important Consideration: Interaction with Simultaneous Evaluations

The "maximum number of concurrent jobs” works with the "maximum number of simultaneous evaluations" setting. To optimize performance, consider the following:

  • If the "maximum number of simultaneous evaluations" is configured to exceed the "maximum number of concurrent jobs," the Power BI engine will constrain parallel processing to the limit defined by the "maximum number of concurrent jobs."

  • If the "maximum number of simultaneous evaluations" is set lower than the "maximum number of concurrent jobs," the engine may initiate the number of concurrent jobs specified (subject to system resource availability). In this case, the "maximum number of simultaneous evaluations" may limit the concurrency of Power Query operations executed by those jobs, resulting in queuing of some operations.

  • It's crucial to note that the evaluation of calculated columns, calculated tables, and other Power BI engine components that do not rely on Power Query is not governed by the "maximum number of simultaneous evaluations" setting. These operations can still leverage the full degree of parallelism allowed by the "maximum number of concurrent jobs”.

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Copyright Kyvos, Inc. 2026. All rights reserved.