Impact of report level filter on capacity utilization

[alohaes]

Scenario:

I got a non-paginated report that I am exporting via the REST api with some report level filters. The exports are done in a loop with each unique filter applied to that export. For testing, the loop increment is limited till 4. The dataset model is in import mode and there is quite a lot of data in the source db. 

Issue:

The issue that I am facing is that when the exports happen, the dataset workload of Azure metrics shows utilization of more than 150% percent but the InteractiveReportExport workload shows almost nil utilization, like 3-4% at times.

Question:

So, I would like to understand whether the filters that I am applying while exporting are causing a massive spike in the dataset workload maybe because the model is not optimized or is there some reason for this? (Below I have also attached a snapshot of dataset utilization when I tested). Really appreciate all the help, thank you!

[Anonymous]

Hi @alohaes ,

Do you mean that you add the filter in your rest api to export the report?

Since you're using Import's connection mode, when you filter the entire REPORT to perform the export operation, the resources in the capacity will be allocated based on the size of your data source. If the data size is large, it will take more resources and show increased occupancy on workflow.

You can do filter in your report firstly or in Power Query Editor. You can also try incremental refresh if you report contains date. This will optmize the Power BI performance.
For reference: Optimization guide for Power BI - Power BI | Microsoft Learn

Best Regards,
Rico Zhou

If this post helps, then please consider Accept it as the solution to help the other members find it more quickly.

[alohaes]

Hi @Anonymous thanks for the reply. Yes, I am applying report level filter while exporting the report. I don't think the report has incremental refresh created so I will try that one out. Though I do have another question. Are the export workflow and sematic model workflow connected somehow when exporting? Like while exporting I believe an interactive refresh happens on the visual side but am not sure if it does in background side as well. If so, it would make sense to create an incremental refresh like you said 🙂

[lbendlin]

Exporting does not create background activities. It is purely interactive.