Single vs Multiple Dataset

[Anonymous]

Hi Expert,

in my company on PBI Service I can see tons of different workspaces with many reports, now almost all reports have a dedicated dataset.

Since there are a lot of dataset for which the original source is a single SSAS Tabular Model, I was wondering if these way of managing the workspaces is correct, what are the differences between one single Dataset as a source for many reports or one dataset for each single reports,  and what pros and cons.

Thanks in advance,

Gio

[v-alq-msft]

Hi, @Anonymous 

Here are the relationships with datasets, report and dashboards. You can also refer to the official document .

ONE dataset...

• can be used over and over in one or in many workspaces.

• can be used in many different reports.

• Visualizations from that one dataset can display on many different dashboards.

ONE report...

• is contained in a single workspace.

• can be associated with multiple dashboards within that workspace. Tiles pinned from that one report can appear on multiple dashboards.

• can be created using data from one dataset. Power BI Desktop can combine more than one data source into a single dataset in a report, and that report can be imported into Power BI.

ONE dashboard...

• is associated with a single workspace

• can display visualizations from many different datasets

• can display visualizations from many different reports

• can display visualizations pinned from other tools (for example, Excel)

I'd like to suggest you use dataflows if you are work with complex or larger projects. They're great for preparing data for use—and potential re-use—by your datasets. For more information, see the Self-service data prep in Power BI article and composite-models . 

Best Regards

Allan

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

[Greg_Deckler]

This one comes down to governance and such. Many people start out with 1 dataset = 1 report = 1 dashboard = 1 app. As you are probably seeing, this can lead to quite a few inefficiencies, especially if many reports are really use the same source systems and data under the hood. If you leave this state of affairs go for very long, you will have a massive pile of spaghetti at some point that you will need to go in and clean up. 

It is better to get a handle on it sooner rather than later. Often, it is far better to create a kind of "golden" dataset that has been professionally modeled and cleaned up and prepared for self-service. Create this dataset without any reports. Then use this dataset via a Live connection or in the Service to create reports. More efficient, one place to maintain the data/data model versus 50. One refresh versus 50, etc.

It is also often a preference to go the Dataflows route to create reuseable queries that other people can use. Again, more efficient. Instead of the same query hiding in 50 different reports, 1 query that everyone uses.

Anyway, don't want to get on my high horse regarding governance but I recommend that you pay attention to it and choose a model that is right for your business. There is no single answer that is 100% "right" all of the time with this stuff.