Microsoft Fabric Tutorial: How to Create a Lakehouse & Shortcuts
This video walks you through some basics of Microsoft Fabric. I’ll show you how to create a lakehouse in Fabric and how to link to external data sources using an ingestion method called shortcuts. After that I’ll demonstrate how to create a semantic model and finally, how to connect to the semantic model and create visuals in Power BI. The tone of these videos is in the same conversational language that helped me better understand Fabric.
If you are skilled in Python and SQL, you may be aware of other ways to accomplish some of these tasks, and that’s ok. This video is really aimed at someone with basic coding knowledge but is not a developer. This type of person may have ideas but can be limited by technical constraints. By the end, you’ll have exposure to some Fabric tools that enable data transformations/manipulation without coding expertise. It also provides useful context for our Fabric Analyst in A Day Workshops to help you get ahead.
The sections below will summarize key concepts of the video. Click the timestamps to jump to the section of the video for step-by-step instructions for each component.
First, let’s cover some helpful definitions of terms in Fabric I use throughout the video.
Microsoft Fabric Basics: What is a Lakehouse? (1:21-3:07)
A lakehouse is a storage layer that can handle structured and unstructured data. This means you can upload any data file to Fabric in various formats.
Structured data can be brought directly into a lakehouse using multiple ingestion methods. Once ingested, it can be used (almost immediately) in a semantic model.
Unstructured data can also be brought into a lakehouse. If you have a messy matrix-format spreadsheet, you can import it and use Fabric's tools to convert it to a data table, process it as delta format, and then use it for analytics.
I don't know lots of Python or how to break down data strictly using code. In Fabric, I can easily take unstructured data, use its ingestion methods and transformation, and end up with structured data that I can use to build some beautiful Power BI reports.
What Is a Fabric Shortcut? (3:06-4:37)
The shortcut feature in Fabric is a data “ingestion” method. If you have data saved in AWS, Google Cloud or even Azure Storage Gen2, you may not want lots of people duplicating these large data tables. Shortcuts create a reference to data, giving people access to large data sets to incorporate into their analysis/reporting without duplicating it.
Now that we’ve covered some basic terminology, let’s create our first Fabric workspace.
Creating a Workspace in Microsoft Fabric (4:38-6:33)
The easiest way to get to Fabric is to Navigate to app.powerbi.com. You’ll need to create a Fabric capacity workspace to leverage lakehouses to ingest data using a shortcut. To confirm that it's a Fabric capacity workspace, go to workspace settings and in license info you need to confirm that the license configuration says: “Fabric capacity” and “Large semantic model storage format.”
Using Fabric incurs cloud compute costs. You might not have permission to update Fabric capacity, and your organization likely restricts this access. Your administrator can help answer these questions.
If you’re ever curious about how Power BI licensing works in Fabric, check out our blog: Your Guide to Power BI Licensing in the Microsoft Fabric Era | OmniData™.
Creating a New Lakehouse in Fabric (6:33)
Creating a lakehouse in Fabric is surprisingly simple.
From your workspace click --> New item. -->Lakehouse --> Give it a name --> Create
Here, I have my favorites selected, but there are lots of different options for actions like: visualizing data, getting data, storing data, preparing data. I only use some of them, so I just favorite them, and then when I log in, it just shows me the things that I care about.
The 2 Components of a Lakehouse (7:10-8:20)
Once inside the lakehouse, there are two components to the lakehouse I highlight. One of them is the lakehouse itself. In here you can view your data, ingest data, create semantic models, and leverage notebooks. (like Databricks).
You also have a SQL analytics endpoint. You can run SQL queries and get a connection string that'll allow you to connect to 3rd party applications, like Tabular Editor or Sequel Server Manager Management Studio (SSMS.)
How to Ingest Data into a Fabric Lakehouse (8:21)
Now that you have a lakehouse, you can bring in data. There are several ways to ingest data in Fabric:
Upload File
If you have a one-off instance that you don’t need a great process around managing, use the upload file section.
Data Pipelines
Data Pipelines work like Azure Data Factory Pipelines.
Dataflow Gen2
Dataflow Gen2 is like the Power Query editor you’ve likely used if you ever ingested data into a PBIX file. The main difference is that in Fabric you have the flexibility of working in the cloud. And you can choose where your data lands when you're done transforming it.
Eventstream
If you need near real time data, use the Eventstream feature in Fabric. For more examples check out OmniAnalytics, our own proprietary real-time analytics for Dynamics 365 ERP built in Fabric.
Shortcuts: How to Create a Shortcut in Microsoft Fabric (9:50-13:45)
A shortcut in Fabric is a data ingestion method that creates a reference to the original source data without duplicating it. It’s useful for external data stored in AWS, Google Cloud, or Azure Storage Gen2, where duplicating large data tables can be inefficient. Shortcuts allow you to incorporate large datasets into your work without creating multiple copies.
I use an internal source and navigate to the workspace that contains the lakehouse I want to retrieve data from.
If your organization uses a medallion structure for their data, there may be multiple lakehouses that appear to house the same data. If you are using the data you ingest to create reports or visuals (as opposed to exploring more ‘raw’ data), always use the curated (or gold) layer of data since it's already been shaped/cleaned/processed by data engineers. For more exploratory work, use the raw tables.
In my example I selected the OmniAnalytics table from the curated layer. Then I select Company, Customer, Date, fact sales, and Posted Transactions tables. It’s helpful to rename the data for business analysts’ convenience. This is the first step in preparing data for end users.
You can quickly identify a shortcut in Fabric by the icon with little delta table symbol in the bottom right, and the hyperlink in the top left corner.
The hyperlink icon indicates a shortcut, referencing the selected source without containing actual data.
Creating a New Semantic Model in Microsoft Fabric (13:45-15:17)
Once you’ve ingested your data into a lakehouse, you are ready to create a semantic model. If you had unstructured data like a messy Excel spreadsheet, you could import it into your files section of the lake house and create a new Data flow Gen2 to transform it and publish it as a structured table in your lakehouse. Fabric provides tools to access, cleanse, and shape both structured and unstructured data.
To create a semantic model: click New Sematic Model --> Name --> Select tables --> click Confirm.
Creating Relationships in a Semantic Model (15:19)
Once the model is created, you can now create relationships, linking your fact table to your dimension tables.
In my example, I connected the date key with the document date key. The keys in the other dimension tables are related in the same way. Now, we have a refined model.
From the semantic model screen, you can also add low level security (RLS) by clicking “Manage Roles” to allow you to control access at a more granular level.
If you didn’t bring in the right tables, go to Edit tables on the semantic model window and select or unselect them from there.
In my example I brought in data from one location via shortcut. You can bring in data from multiple locations via shortcut. You can create shortcuts to tables from various lakehouses. Then use your modelling ability and the transformation tools in Fabric to tie those together.
Creating a Power BI Report from a Semantic Model in Fabric (17:22-19:46)
Now that we have a model, and have refreshed it, you can create a report.
Go back your workspace --> click New item in your workspace --> choose Report.
You can create a report from the semantic model you made using data from another lakehouse, without duplicating any data. This keeps your data footprint small. You can publish the report and store it in a folder if needed.
When you return to your workspace, you'll see your lakehouse, SQL analytics endpoint, semantic model, and the report created from that model. The lineage will show how everything connects.
By default, every lakehouse includes a semantic model. While I didn't explore all its uses, this video's focus was on creating reports, semantic models, lakehouses, and shortcuts.
Next Steps Video
This is the first part of the Microsoft Fabric Basics tutorial video series. In part 2, I will explain how to use the Data Flow Gen2 tool to clean an Excel budget file for integration with other data.
For more hands-on experience with Microsoft Fabric and Power BI, register for one of our free twice-monthly workshops:
