Forum Discussion
PowerBI Semantic model Refresh taking long time
Hello,
I have run into an issue in our PowerBI Service. We have a lot of semantic models and they get refreshed on a schedule. While monitoring some semantic models take over 3-4 hours while.
If we try and refresh the same model in PowerBI Desktop it completes the refresh in 30 sec. I am wondering how can i check why model refresh takes 10X time on PowerBI Service and how to avoid this bottleneck. It does show its processing but never hits the database when we have trace running on the database.
Appreciate any guidance on this, please let me know if more details are needed.
there are could be numerous reasons, for example:
- it could be your Data gateway resources if you use one for refresh,- could be capacity/Power BI service busy with other task (you have compelx transformations in Power Query or load tones of data that need to be compressed)
Take a look at this article, using this aproach I track how much it takes to load and process data and identify slowest queries to optimize:
Visualize Power BI Refresh using Semantic-link - Phil Seamark on DAXHi Ira_27 ,
Thank you for engaging with the Microsoft Fabric Community. Here are a few quick checks to help identify where time is being spent.
1. Download the refresh history for the dataset and review how time is split between Data Load and Processing.
2. For Premium/PPU, use the Capacity Metrics app to check CPU, memory usage, and any throttling during the refresh period.3. Make sure that intensive Power Query steps are folding to the source; if not, the Service handles more processing.
4. Large single partitions and columns with high cardinality or width can increase processing time. Partition large fact tables by date and remove unused columns.
5. Gateway (if used)Check CPU/memory and queued requests on the gateway. High activity can slow things down before source queries run.
For a quick improvement, enable Incremental Refresh to process only recent data instead of the entire history. This change often has the greatest impact on large models.
Helpful Reference: Incremental refresh for semantic models in Power BI - Power BI | Microsoft Learn
-Yugandhar.
5 Replies
there are could be numerous reasons, for example:
- it could be your Data gateway resources if you use one for refresh,- could be capacity/Power BI service busy with other task (you have compelx transformations in Power Query or load tones of data that need to be compressed)
Take a look at this article, using this aproach I track how much it takes to load and process data and identify slowest queries to optimize:
Visualize Power BI Refresh using Semantic-link - Phil Seamark on DAXHi Ira_27 ,
Thank you for engaging with the Microsoft Fabric Community. Here are a few quick checks to help identify where time is being spent.
1. Download the refresh history for the dataset and review how time is split between Data Load and Processing.
2. For Premium/PPU, use the Capacity Metrics app to check CPU, memory usage, and any throttling during the refresh period.3. Make sure that intensive Power Query steps are folding to the source; if not, the Service handles more processing.
4. Large single partitions and columns with high cardinality or width can increase processing time. Partition large fact tables by date and remove unused columns.
5. Gateway (if used)Check CPU/memory and queued requests on the gateway. High activity can slow things down before source queries run.
For a quick improvement, enable Incremental Refresh to process only recent data instead of the entire history. This change often has the greatest impact on large models.
Helpful Reference: Incremental refresh for semantic models in Power BI - Power BI | Microsoft Learn
-Yugandhar.Hi Ira_27 ,
Could you let us know if your issue has been resolved or if you are still experiencing difficulties? Your feedback is valuable to the community and can help others facing similar problems.Hi Ira_27 ,
Has your issue been resolved, or do you still need assistance. If you need anything else or more information, please let us know. We're here to help.
Thank you.
Hi Ira_27 ,
May I know if your issue is resolved, or if you still need any additional help? Please let us know.Thank you