Query Regarding PBI Dataflows

[srlabhe123]

Hi All,

As I am delaing with large datasets, after reserach I found that the PBI Dataflows is best suitable option for same.

I have designed a simple dataflow with defing bew tables form Oracls database. I also designed PBI dastset from the same datflow.

Oracle Databse->PBI Dataflow-> PBI Dataset (Direct Query)-> PBI Report

Now with that I have few questions as below:

1. If we wanted to do transformations like Renaming column, filtering rows or defining measures , is that suggested to do on Dataflow level or PBI dataset level. Whats standrad practice here?

2. How to make sure those transformations to reflect across Fron PBI Dataflow to all the way to PBI report? Just refresh would work?

3. Whenever I try transformation on PBI dataflow level say on Table1, and try to pull that in  PBI Dataset (Direct Query) it shows the Table1 has transformations and cant be pulled as Direct Query.Here the main purpose of creating direct query PBI dataset goes waste. Is that right behaviour?

Appreciate your suggetsions here please.

[v-henryk-mstf]

Hi @srlabhe123 ,

Dataflow entities that support direct query display the "view" icon in the Power BI desktop instead of the "table" icon, so no other transformations can be performed and can only be changed on the original data source.

To use DirectQuery with dataflows, you must explicitly toggle the enhanced compute engine to On in dataflow settings. You must then refresh the dataflow before it can be consumed in DirectQuery mode.

Direct query is different from import mode, and the above steps must be carried out.

For more details, you can read related document as below:

Using DirectQuery with dataflows - Power BI | Microsoft Docs

If the problem is still not resolved, please provide detailed error information and let me know immediately. Looking forward to your reply.

Best Regards,
Henry

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

[lbendlin]

Direct Query against dataflows is a sham.  As Microsoft explained at MBAS what really happens is that they put an Azure SQL database in between the dataflows and the datasets, and then do the direct query against that database.  This completely defeats the idea of a dataflow (which is similar to csv or parquet blob storage)

1. Standard practice is to do transforms as high upstream as possible, but not higher! Every transform tampers with the original data and has the potential to destroy information.

2. that is automatic

3. Don't use Direct Query.

[srlabhe123]

Thanks for your reply but then idea of Dataflows is not useful

[lbendlin]

They are useful for storing large amounts of tabular-ish data that you can then ingest quickly in import mode. They can also act as a buffer between a very slow source system and your developer experience.