Dataflows in Golden Dataset

[astanfo]

I am working to implement Power BI framework that includes best practices for use of data in Power BI reports throughout the organization. From my research, I think using a Golden Dataset is the way to go. I also think using Dataflows can be beneficial, so analysts can create reports using the upkept dataflows as a source of truth instead of having to recreate connections to SharePoint folders (unfortunately we cannot connect directly to SAP HANA so flat files/workbooks hosted on SharePoint is a must). Trying to keep an agile mindset and make everything as simple as possible for analysts to grab data to create analysis.

Since I would be looking to use Dataflows AND a Golden Dataset, I see the benefit of using dataflows within the golden dataset. HOWEVER, I would like the data to get updated as quickly as possible in the golden dataset (Not 45+ minutes), and I do not believe I can create incremental refreshes for the dataflows AND the golden dataset. The idea would be to refresh flat files on SharePoint, which would then kick off a Power Automate flow to refresh the dataflows and golden dataset.

My Questions:

1. Is it best practice to connect to dataflows in a golden dataset or connect directly to the source within the golden dataset so that I can use incremental refresh for fast refreshes in the golden dataset and the dataflows?

2. Would it be better/more efficient to not focus on dataflows and just have analysts get data from the SharePoint folder instead? I don't want to overcomplicate things by having to maintain dataflows if it is not beneficial. Curious what others experience is on this. I have come across people having problems with dataflows, and if I don't need to use them and deal with dataflow problems then maybe I should not use dataflows.

[TomMartens · Accepted solution]

Hey @astanfo ,

using dataflows and datasets are basically not the same thing, of course 🙂

A golden dataset does not need to be using dataflows as a data source. A golden dataset removes the burden from its users of

• extracting data from various data sources
• data modeling and DAX challenges.

Personally, I consider golden dataflows as an (agreed upon) representation of business objects that can be reused throughout various datasets in your organization.

So, depending on what you want to provide also in regards to the most current data, I tend to provide both

• a dataset and
• dataflows

but the dataset does not use the dataflows, instead, it directly connects to the data source.

Dataflows are a great way to remove the burden of extracting/cleaning/transforming from analysts.

Hopefully, this answer helps to make your decision.

Regards,

Tom

[astanfo]

Thanks @TomMartens! Very helpful.