Exploring Microsoft Fabric from a Power BI Perspective

You’ve probably heard or read about Microsoft Fabric, at least if you’re into data and analytics.

This blog post will cover common questions and personal thoughts through the lens of Power BI.

Azure Synapse, Power BI, and Data Factory are all integrated into Microsoft Fabric. The purpose of uniting these components is to simplify the collaboration between Data Engineers, Data Analysts, and Data Scientists. The below picture describes the components that are included.

An assumption is that it’s just a combination of products. But let’s take a stroll down memory lane to June 2015. Microsoft collected add-ins from Excel; Power Pivot, Power View, and Power Query and created what we today call Power BI. With monthly updates for the desktop and weekly improvements in the service, it has grown to become the leader. I believe that this success comes from one key point – keeping things simple. Microsoft is investing in Fabric, I therefore think that it’s going to turn out to be something more than just a collection of tools.

You don’t have to migrate your workspaces or content to Fabric since Fabric comes with the same look and feel as the Power BI portal.

Where do I store the data?

OneLake is where Fabric stores all its data, like OneDrive for your files. OneLake is included with each setup of Fabric tenant and is accessible through Windows File Explorer. Accessing workspaces, files, and downloading data are just a few examples of tasks that are possible.  

The purpose of OneLake is to maintain one single copy of the data, eliminating the need for data transportation and duplication. Data is stored in parquet format, which means that data is compressed and therefore enables better performance.

But what does this mean in terms of Power BI?

I believe that many Power BI developers have faced challenges related to version control, collaboration, performance or just getting an overview of how the artifacts depend on each other. 

Luckily, Fabric comes with a lot of new and exciting features!

Direct Lake  

This is the latest addition to Power BI connection modes, joining import, direct query, live connection, and composite model. With its connection to OneLake and parquet files, it operates faster than direct query.

The direct query connector is built to consume real-time data. It doesn’t duplicate data like the import mode, but it sends queries to the database, resulting in slower performance.

The import mode is more efficient because it involves copying and caching data, but it does require scheduled refreshes and has a limitation of 1GB dataset.

If you combine the strengths of these connection modes – fast data access and non-duplication, you reach Direct Lake. As this feature is still in preview it comes with certain limitations, for example, no support for calculated columns/tables, and it’s not possible to combine other connection modes with direct lake, i.e., creating composite models.

Direct Query is perfect in scenarios where you have datasets that don’t fit into memory or need to analyze real-time data. The below picture illustrates a clearer view.

Power BI Project (.PBIP)

This is a new format in Power BI Desktop that enables you to save reports and datasets in plain text files. Text files provide several possibilities that improve, not only report development but also collaboration and authoring. A project folder is created containing reports and a dataset. Each report is saved in the same project structure as shown below.

This presents the opportunity for modification in tools like VS code, Tabular Editor, or Notepad.

The following files are not editable since the feature is currently in preview.

• Report\
• report.json
• mobileState.json
• datasetDiagramLayout.json
• Dataset\
• diagramLayout.json

Keep in mind that Power BI Desktop needs a restart to reflect changes that have been made in an open-source tool. While in preview, there are other limitations that you should consider. Publishing to the cloud is for example disabled while working in PBIP files. To deploy to the cloud, save it as a PBIX file or use the fabric git integration. Downloading PBIX files from the service is unavailable if the workspace is integrated with Git.

Git Integration

This is likely one of the most requested features! It’s integrated into Fabric and Power BI, making it compatible with Power BI workspaces as well as Fabric workspaces. You’ll find the Git integration option in the workspace settings.

To get started with this, you’ll have to create a repository for your team in Azure DevOps and then sync it with your local repo in VS Code. The local repo is your PBIP file structure mentioned earlier in this post.

This enables better collaboration in terms of allowing multiple developers to work on the same Power BI project. The picture below presents the overall architecture.

The Git integration feature is only supported in Premium.

What about licensing?

- Capacity license is like the Premium version in Power BI. It provides access to all the components and features in Fabric. This license is divided into SKUs (stock-keeping units) which means that depending on the SKU, you get different sets of Fabric recourses.

How does the Power BI licensing affect Fabric?

The Power BI licenses are still present but with minor changes. The below list of licenses applies to workspaces located in the Fabric platform.

- Premium Per User (PPU) only gives access to Power BI items. It also provides partial access to Fabric but at a lower cost. This option is well-suited for smaller companies.

- Pro licenses allow for creating and sharing Fabric content. Pro license is required if you intend to work as a developer.

- Free licenses give users access to their own content in “My Workspace”. In other words, if you need to access the portal and play around, this is the license to go for. However, if you are a free user and your organization has a Fabric capacity, then users can consume content, given that the workspace is backed by at least F64 which equals to P1 or A4 SKU. If you have a lower SKU, users require a Pro license to use Power BI content.

The following picture illustrates the translation of Fabric units into Power BI SKUs.

Microsoft Fabric will be updated monthly, releasing new and improved functions. I believe that enterprise companies that struggle with silos may be at the forefront of testing Fabric. As with new products, it will take some time to adapt, not so much for Power BI developers but for Data engineers and Data Scientists who are unfamiliar with the portal. I think there will be a lot of questions and discussions around security, domain setup, and how it will work from a high-level perspective. Lastly, I think that the release of the new tool Data Activator will be a hot topic.