If you run Power BI, you may have come across references to "Fabric". At least you have seen this logo when logging into the Power BI Service for the last few months. What is it? And if all you want to use is Power BI, why should you care?
With the Microsoft Fabric Community Conference just two weeks away, it's a good time to ask this question. This article covers what Fabric is and what it means for your business. In this first part of the 2-part series, we will explore the potential advantages of Microsoft Fabric for you and your business. In Part 2, we will explore important cautions you need to know if you opt to build on Fabric.
It's Power BI + Data Engineering + Data Science tools. Specifically, Fabric a suite of tools and a cloud services that make data engineering and data science workloads run alongside Power BI on the same cloud platform.
Power BI has always come with the ability to move data using PowerQuery. However, PowerQuery was designed specifically for citizen ETL, and although it quickly and easily connects to hundreds of data sources, it's available data sinks are quite limited (basically tabular models or ADLS). While quite useful for the data analyst, its uses as a professional data engineering tool are limited. Enterprise-scale analytics and data workloads require more robust tools to achieve their strategic goals; tools like lakehouses, notebooks, data pipelines, and databases. In the past, if you needed these tools, you had to build your own using Microsoft Azure resources or some other platform.
Microsoft's vision with Fabric is to bring all of these tools, along with Power BI, under a "single pane of glass" and allow you to easily "stitch" all your data together into one cohesive surface (thus the name Fabric.) The success of Power BI is largely due to the success of the Power BI cloud services that Microsoft has built under them for the past 10 years; while cloud was an afterthought for some BI vendors, Power BI was always cloud native. Fabric is the expansion of this platform to host the entire data-for-analytics lifecycle.
No! If you purely want to focus on using Power BI as you have always done, that is a valid option*.
Still when Microsoft invests so heavily in something, it's often worth asking what benefits may be in it for your business. Here are 5 possible advantages that Fabric may bring to businesses already running Power BI.
Fabric SKUs are available at lower entry points than Power BI Premium. The Premium P1 capacity with 8 v-cores is equivalent to an F64 Fabric capacity. Below this level, Fabric capacities are available all as F2, F4, F8, F16, and F32 sizes. So, if you are a company running Pro or PPU, you can begin to explore all of these features all the way down at the F2 level, starting as low as $156/mo with a reserved capacity. This is a great way to get started with the technology to see if it may help with your data workloads.
If you are already running Power BI Premium or Power BI Embedded, these smaller capacities give you lower-priced options that weren't available before.
For companies who are running data workloads in a pay-as-you-go model in Azure, Fabric can also offer cost simplifications. You purchase a capacity at a know cost, then run whatever workloads you need on them. The costs tend to be more predictable this way.
Note that there are some Power BI licensing differences above and below the F64 level and capacity management is a new way of thinking; we will cover those in part 2.
If data workloads currently run as native objects in Azure or AWS, you may find some management challenges. First keeping these workloads organized by their purpose and business-criticality can be daunting. Second, cloud platforms and their resources tend to require a developer-level mindset and expertise to change. Finally, knowing how to allocate the cost of the various workloads can also be confusing, relying on a sophisticated system of resource groups, tags, and naming conventions (not to mention deciphering the Azure bill!)
Fabric can help greatly simplify this by providing dedicated capacities for various departments or consumer groups within your organization. Additionally, the tooling is all browser based and will tend to appear in the same familiar experience that business users already have with Power BI. This makes citizen ETL more achievable. As various groups' workloads are isolated to their respective capacities, you can give them all more freedom to experiment using all the resources that you have made available to them.
Fabric has also simplified the naming of these objects and reduced the choices available to the users, which provides much more clarity to their experience. For example, when creating a new object in Fabric and searching for "data", you are currently presented with about a dozen options including "dataflow", "data pipeline" and "SQL database". A similar search in Azure yields 11,100 results. Some simple conventions should lead to a more unified data experience for both citizen ETL developers and IT pros.
Underlying all of Fabric, and included when you create your first Fabric capacity, is a delta-parquet format data lake that is native, and automatically provisioned, for all your data in Fabric to use. Much of the effort of bringing objects from Azure to Fabric consists of teaching all Fabric objects to natively read and write from this new format. Even Power BI semantic models can natively read from and write to this format.
This innovation opens up a new kind of connection for Power BI. In addition to Import and DirectQuery, you now have DirectLake as a connectivity option. DirectLake enables Power BI to read data directly from the delta-parquet format data lake in Fabric without any intermediate import or caching steps. This can potentially enhance performance by reducing latency and is especially useful in real-time reporting scenarios. You can always rely on Patrick Leblanc and Guy in a Cube to dive headfirst into the OneLake!
Fabric introduces many new types of objects and workloads that can run side-by-side with Power BI. Two, we think, are especially useful for Power BI shops.
Fabric Data Activator can take actions automatically by monitoring Power BI data and calculation results. Users can create and manage their own alerts, setting up much more sophisticated conditions including rolling windows. This takes alerts to the next level for end users to be alerted when data triggers certain conditions. They also have many options related to how they choose to be notified.
If you work with data every day and are interested in learning to code, Notebooks are an excellent tool to get started and run code in a very user-friendly and interactive way, with little friction regarding setting up a development environment.
Where these could especially benefit Power BI shops is their reusability. For example, here are some great functions you can do in Power BI with Fabric notebooks, including getting the Power BI semantic model size, listing out capacities, and even rebinding Power BI reports to different semantic models.
This is a catch-all, but Fabric is the only system that is fully integrated with the Microsoft Entra ID, Microsoft Office, and Microsoft Purview Information Protection.
Even if Fabric does not become your central repository for all corporate data, it makes a strong use case to become your "platinum medallion layer" or your "last mile data provider". You can virtually include other data sources in it through replication and direct query. Also, the starting point of a data journey for many users is the data visualization layer. It makes sense for them to start at Power BI, then trace its lineage back to the intermediate data objects and ultimately the data sources as they increase in data literacy.
Fabric's value promise is to serve as a unified reporting and analytics layer for business users. Going back to its interwoven namesake, it wants to be the place where all your data analytics work finally comes together, whether that work is done on the desktop of a single analyst, in the notebooks of your data scientists, or as enterprise data models by your professional data engineers. Piethein Strengholt gives some great examples of how Fabric might enhance even a robust data platform like Databricks in this article.
If you have not yet signed up for the 2025 Fabric Conference March 31st - April 2nd in Las Vegas there are tickets available - and we have discount codes! Just email info@firstlightanalytics to get one of those. This is the big Power BI conference of the year. We will be there sponsoring booth 116, and we would love to meet you in person and talk through specific questions you have about Power BI or Fabric.