Data Flow refresh - parallel or separately

[Anonymous]

Hello,

I was exploring the Power BI Data Flow service and it looks very promising.

However, I couldn't find any information about how the Data Flow Service refreshes it's entities. 

In Power BI Desktop it is possible to enable or disable parallel loading of tables from the settings. I couldn't find this kind of option from the Data Flow Service or any documentation of how the entities are refreshed. 

In some situations we prefer to disable the parallel loading of tables and run the queries separately as our data source (OData service) can sometimes get too much load if a lot of queries are run in parallel. 

My questions are:

Does the Data Flow service load the entities in parallel or separately?

Is it possible to enable or disable parallel loading for entities in Data Flow?

If I understood correctly, with Data Flow Service the data is loaded to Azure Data Lake Storage and then we use this Data Flow service as our data source when creating reports in Power BI Desktop. However the Power BI Desktop's option for parallel loading does not affect the original data source, so this option is not valid for controlling the load on our server if we build the reports with Data Flows. 

[v-yuta-msft · Accepted solution]

@Anonymous ,

Currently there's no such kind of doc which can clarify the working details about dataflow entity refresh. But it seems to more related to Azure Lake Storage. I would suggest you to post this question to azure forum to achieve more help.

Community Support Team _ Jimmy Tao

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

[Anonymous]

I have the same problem where i need to control the order in which the entities in a dataFlow refress in order to avoid API call quota limits on the source side (Google Analytics).

To me this has nothing to do with DataLake store but rather with the way DataFlows decides to treat the refresh calls towards the datasource.

Is there any documentation that would explain how multiple entities in a flow are refreshed (sequentially or parallel)?

Thanks

Bogdan

[Imrans123]

Any update on this issue? I am using Invoke After as a work around as I have multiple tables using the same API source and I'm hitting the quota limit. But it takes a long time to refresh as I give exessive breathing room between queries since sometimes the API works slow..

[Marius0402]

I have the same issue which causes an error when refreshing my dataflow "Parallel API request not permitted". 
Any help would be appreciated. 

[PawelWrona]

You can download the refresh history form Dataflows:

Inside you will find StartTime and EndTime for each entity

As you can see even on my example, some of the are overlapping, this might cause an issue. In this case, probably it's better to split entities to individual dataflows and run them in sequence, if this is your data source limitation.

Also, if you have issue with too frequent API Calls according to your data source rules, you may have a look at the great post from Chris Webb:

Chris Webb's BI Blog: Using Function.InvokeAfter() In Power Query (crossjoin.co.uk)