Power BI Embedded data load optimization

0
votes

I have some issues with the performance of Embedded Power BI reports. I want to make the visuals load faster and I'm wondering how the data load is working the backend with PBI Embedded.

I have two large datasets (100kx20) used in a report. One of them is used in macro-economics sheets and the other is used for micro. The Micro part is split into 5 categories which are visualized in 5 different sheets. The question is if I open up 1 of those sheets does the Embedded service load the whole dataset2 and filter it out for 1st category? Or are the datasets loaded into Azure service as soon as the report itself is opened?

The question is would splitting the dataset2 into 5 different sets make the data in those 5 category sheets load faster?

1
powerbipowerbi-embedded

1 Answers

0
votes

In Power BI Embedded, the report/datasets are stored on blob storage in the background. When the report is opened, it will load the report from blob storage and the associated dataset(s) into memory. So when you open the report it will load the dataset then apply the filter. It will remain in memory until it is evicted. Evitions will be for two reasons, it is not longer in use, for example no more interactions with the report, or there is memory pressure, and the Power BI Service needs to allocated the memory to another report(s). Some more details here. It is for premium, but it works roughly the same in the background.

For Power BI Embedded, EM and A SKU's levels 1 and 2 are shared capacity, Only level 3 SKU's and above are dedicated capacity, like Power BI Premium.

One other consideration is the number of visuals that you have in the report, as each visual is a query, so a large number of visuals will take time to query, return, and display the data.

Depending on the size of your embedded instance, it may be beneficial to separate the reports into individual reports to reduce the background load time and data set memory footprint

One good way of looking at whats going on is to have a look in the Power BI Portal and the metrics for your embedded instance. Your can also use the premium metrics app to see whats going on. It does work with Embedded instances as well. I have used it on a Power BI A SKU.

Hope that answers your question