Forum Discussion
Power BI Performance Bench mark
Use the "Power BI Premium Capacity Metrics" template app to collect and investigate your capacity performance. Depending on what P SKU you have you may be able to segregate traffic. For example instead of one P2 SKU you could run two P1 SKUs, one for the well behaving users and one for the users on the nasty list.
Take a very hard look at the additional workloads that you have enabled. Paginated Reports, Dataflows, and AI/ML all try to eat your memory away. Don't let them if there is no ROI.
Create a schedule refresh heatmap. Most users do not understand the half hour concept, and everybody schedules their reports at 10am and 12pm. Educate your users on API based refresh.
Monitor refresh frequency and (much more important) cumulative query duration per dataset. Work with the developers that are hitting your capacity the most to educate them on best practices how to reduce their impact on the capacity.
Offer lots of training to developers. Show them the Premium capacity metrics often. Stay on top of eviction counts, memory starvation and renderer scarcity.
Hi, Anonymous
Power BI Performance is affected by many factors like capacity settings, network latency and so on. I'd like to suggest you refer to the document to optimize performance for Power BI.
Best Regards
Allan
If this post helps, then please consider Accept it as the solution to help the other members find it more quickly.
Use the "Power BI Premium Capacity Metrics" template app to collect and investigate your capacity performance. Depending on what P SKU you have you may be able to segregate traffic. For example instead of one P2 SKU you could run two P1 SKUs, one for the well behaving users and one for the users on the nasty list.
Take a very hard look at the additional workloads that you have enabled. Paginated Reports, Dataflows, and AI/ML all try to eat your memory away. Don't let them if there is no ROI.
Create a schedule refresh heatmap. Most users do not understand the half hour concept, and everybody schedules their reports at 10am and 12pm. Educate your users on API based refresh.
Monitor refresh frequency and (much more important) cumulative query duration per dataset. Work with the developers that are hitting your capacity the most to educate them on best practices how to reduce their impact on the capacity.
Offer lots of training to developers. Show them the Premium capacity metrics often. Stay on top of eviction counts, memory starvation and renderer scarcity.