Forum Discussion
Power BI Slow Report Performance - Service
Have a look into the capacity metric, see if you can identify any processes that take significantly long or it could also be throttling on your capacity if it hits the limit?
Hi TotalAnonymous - Your users are in Japan, and the Power BI Service data is stored in Ireland. The geographical distance could introduce latency when users interact with reports.
Recommendation: Consider moving your dataset storage region closer to where your users are based (Japan) to minimize latency. Alternatively, using Power BI Premium capacities with multi-geo options could allow for closer storage to users.
The time it takes to load a report isn't only dependent on query performance. Complex visuals (or too many visuals) on a page can slow down rendering.
Recommendation: Optimize your visuals by reducing the number of charts or using lighter visuals. You can also look into aggregating data at higher levels to reduce the rendering load.
Use Performance Analyzer in Power BI to check which visuals take the longest to render and see if simplifying them improves performance.
Review your gateway cluster and settings. Consider whether scaling up the gateway capacity or adding more nodes could help.
Ensure your gateway is located as close as possible to your data source for quicker data refreshes and lower response times.
Where possible, replace custom visuals with native visuals in Power BI. If custom visuals are required, try to minimize their use or find optimized versions.
Vertipaq Analyzer to ensure that your DAX measures are highly efficient. Complex calculations can be offloaded using calculated columns in your model where appropriate.
Hope the above suggestions helps in optimization.
Thanks rajendraongole1 , for your feedback!
As I mentioned, I've tested my DAX Query for "table creation" using DAX studio by clearing the cache as shown in the image above. The largest number of duration time came from the Table. That's why I confuse our users have taken long while (around 15-40 seconds) to load the table. For the visuals, I use the native rather than the custom one. On that page, I have 1 Slicer, 2 text boxes, 1 bookmark navigator, and 1 table. So, I'm not sure if it necessary to reduce number of visuals again 🙂
On that table, I have 9 fields. 2 among them are column field from dimension table. While the rest are measures.
Our workspace also using premium capacities as shown in the image below.
Furthermore, your recommendation about dataset storage movement and Gateway Cluster are interesting. Have you ever done it before? are they difficult to be done?Have a look into the capacity metric, see if you can identify any processes that take significantly long or it could also be throttling on your capacity if it hits the limit?
- Thanks Tutu_in_YYC, for your reply! Since I'm not a capacity admin, I tried to reach out someone who has access to capacity metric to see how our workspace behave. I've been showed the graph and it seems that the capacity is constantly overloaded. Lately, Background% achieve the %CU limit. Since our workspace is assigned to the regional capacity, i don't know how to solve the issue. Because scaling up the capacity requires approval from higher level.
