Microsoft Fabric - The analytics platform built for the era of AI

Explore Fabric, Microsoft's unified data analytics platform. Learn about OneLake, AI integration, and why it's a game-changer in the field.

Unveiled at the Microsoft Build 2023, Microsoft Fabric is a revolutionary unified platform designed to streamline data analytics. Its unique feature is a unified capacity model, which optimizes computing power across workloads, reducing costs and simplifying resource management. Fabric represents an evolution from existing Microsoft analytics solutions towards a simplified Software as a Service (SaaS) offering, allowing for a smooth transition to this innovative platform. Currently, Fabric is in its public preview stage, with a free trial available for users.

What is Microsoft Fabric?

Over the past two years, Microsoft has undertaken a complete reimagining of the data analytics workflow, resulting in the creation of Fabric. Fabric represents the seamless integration of Azure Data Factory, Synapse Analytics, and Power BI into a unified platform delivered as Software as a Service (SaaS). Fabric's mission is to be "The Analytics Platform Built for the Era of AI." Fabric covers the entire analytics process, from data ingestion to transformation to analysis and presentation. Thanks to its SaaS foundation, everything works smoothly and is automatically optimized, ensuring a streamlined experience for developers and business users!

The concept of "oneness" is fundamental to the platform. With Fabric, there will be one shared compute cost per capacity, one shared governance model across products, one space for data professionals and business users, and more. All this unity between products sits on top of the new OneLake.

Why is OneLake important?

OneLake is being considered the "OneDrive for Data." Like OneDrive, every organization will have one and only one OneLake. OneLake is meant to be efficient and secure while supplying a single source of truth. All tabular data is stored as open-sourced Delta Parquet. Delta Parquet wins out in Fabric because Parquet files are incredibly fast to query, and Delta allows write operations into Parquet (ACID transactions).

Every compute engine (T-SQL, Spark, KQL, and Analysis Services / Power BI) can now query against Delta Parquet. That means everyone can be talking to the exact same data! But it doesn't stop there. If your organization uses tools outside of Fabric for analytic workloads on Delta Parquet, you can still use OneLake as the source! Even though compute engines and tabular stores will be held in Delta Parquet by default, OneLake is still a data lake that can store any file type.

OneLake simplifies the entire security model. Before, we would define different permissions in each product or sometimes even recreate existing logic in each product. Now, security is defined only once and used across every engine/individual. OneSecurity will leverage Azure Active Directory and secure data to the row, column, and object levels.

OneLake emphasizes "single source of truth" frameworks. You can now create shortcuts from one workspace's data to another so that the 'truth' never leaves its home and becomes out of sync. This is amazing for cross-functional data stores. But it doesn't stop there! You can even create shortcuts out of OneLake and into other cloud vendors.

Where does AI come into play?

Everywhere! Microsoft announced they are expanding the vision of copilots (natural language assistants, like ChatGPT) into nearly every Microsoft product (including M365 and Windows). So yes, there will eventually be a copilot in every experience on Fabric too. They will come in two main forms – developer and business user copilots. Developers will have a copilot alongside them to build data pipelines, write Jupyter notebooks, build Power BI reports, and so much more. Business users will use copilots to chat with their data in a multi-turn contextual conversation! This is insane!

Should I ditch everything and migrate to Fabric today?

Well, no (at least not yet). For one, it was only announced last week and is still in public preview. Microsoft offers a free trial to start exploring Fabric (with quite a large computing capacity). Definitely take advantage of that! However, if you want to explore new POC use cases where an AI-empowered, unified analytics solution would be the right fit, I would onboard onto Fabric and OneLake. No date is set for general availability, but we should see a public roadmap any day now.

Wrapping things up

There's still so much content for Fabric. I haven't even gotten into all the bells and whistles of the workflow in the product. Or best practices for licensing, governance, and deployment. So stay tuned.

I see a bright future with Microsoft Fabric. In line with a data fabric approach, teams will be empowered to generate and maintain their data and cross-collaborate within an organization. Let's break down data silos across lakes/warehouses and tech stacks. Let's start collaborating more efficiently between individuals and teams. Fabric was built to empower every data professional and every team (no-code, low-code, pro-code). Let's go build.