Dataflows: Enhanced Compute Engine. Optimized or On?

[frithjof_v]

According to this blog post, the Dataflow connector supports query folding, but only if the Enhanced Compute Engine setting on the dataflow is set to "On": 

"2. Dataset refresh operations can also fold to the enhanced compute engine, meaning even incremental refresh from a dataset can fold to a dataflow, improving refresh performance and potentially decreasing latency between refresh cycles. Just make sure the compute engine is explicitly set to “On.”"

https://powerbi.microsoft.com/en-us/blog/power-bi-august-2021-feature-summary/#post-16859-_Toc794195...

See also: https://blog.crossjoin.co.uk/2021/08/22/how-query-folding-and-the-new-power-bi-dataflows-connector-c...

It sounds like semantic model (dataset) refreshes in the Power BI service can become a lot more efficient, if the compute engine is set to "On" in the dataflow settings.

Does anyone have experience with this?

Should we switch the Enhanced Compute Engine setting from "Optimized" (the default setting) to "On"?

I am wondering if this will enable query folding for the refresh of Import Mode semantic models in the Power BI service, when using the dataflow as the data source?"

[GilbertQ]

Hi @frithjof_v 

Yes, the reason you weren't too much of a speed improvement is because it is all working within the power of vi service. And that is why the refresh is so fast, all. certainly all be faster if you enable the enhanced compute engine.

[GilbertQ]

Hi @frithjof_v 

Yes, this will benefit your refreshers because it will only have to get the data that is needed for the incremental refresh instead of having to query the entire data set.

[frithjof_v]

Thanks @GilbertQ ,

If I don't use Incremental Refresh or DirectQuery, instead I just use normal Import Mode, does the Dataflow connector support query folding when the Import mode semantic model refreshes in the Power BI service?

When I query the Dataflow from Power BI Desktop, I see a big speed boost when I turn the Dataflow's Enhanced Compute Engine setting to "On". When the Enhanced Compute Engine setting is "Optimized", then Power BI Desktop uses longer time to query the Dataflow.

However when I publish the same Power BI Semantic Model to the Power BI service, I don't see any noticeable change in refresh time if the setting is "On" or "Optimized".

I am using Import Mode, but I am not using Incremental Refresh.

Does the Dataflow connector support query folding when refreshing an import mode semantic model in the Power BI service?

Thanks!