In this article, I provide a comprehensive overview of everything you need to know before considering a Lakehouse for your team or projects, especially if you are migrating to Microsoft Fabric for the first time or if you plan to migrate to Microsoft Fabric soon.
The Lakehouse is a data storage system built on the unified One Lake Storage backed by Microsoft security and governance.
What is a Lakehouse
A Lakehouse in Microsoft Fabric is a unified platform for storing, managing and analysing data. It combines the capabilities of a data lake and the structure of a data warehouse to create a data architecture.
One of the core features of a Lakehouse is that it has a Tables and Files where structured, semi-structured and unstructured data are in a single location. This distinguishes it from other data storage systems available within Microsoft Fabric.
The Lakehouse architecture is built on the following:
- Storage layer: Uses Open Data Format with Delta.
- The Processing Layer: Supports Data Pipelines, Dataflow and Notebook/Spark jobs.
- The Consumption Layer: This includes Power BI and Machine Learning Models.
1. Tables:
The Tables in a Lakehouse are managed tables (Spark manages both data and metadata). It supports both delta tables and other types of tables, such as parquet and CSV. A delta table in a Lakehouse is usually indicated with a small black triangle.
2. Files:
Files in a Lakehouse are an unmanaged area (Spark handles only metadata) that can store all kinds of data in any file format (CSVs, images, videos, etc). Delta files stored in the Files section of a Lakehouse are not automatically recognised as Tables. This is why data available in the Files must be loaded to a delta table or use a shortcut to point it to the unmanaged folder in the files section in Spark.
SQL Analytics Endpoint of a Lakehouse
The SQL Analytics Endpoint of a Lakehouse has only the Tables in delta format, and you cannot read the Files Section of a Lakehouse here, including Parquet, CSV, and other formats, cannot be queried using the SQL analytics endpoint. It has a read-only view that allows you to use T-SQL to query data. It does not offer robust T-SQL capabilities, unlike a transactional warehouse and database. It only allows Full Data Query Language, No Data Manipulation Language, limited Data Definition Language and Views.
When should you prefer Lakehouse over other types of Data Lake — A decision Guide.
Choosing a Storage layer for your fabric workload depends on several factors, including the context of your data management, the current skill level of the team, different data types, existing solutions, and other use cases.
In most cases, you use a combination of other data storage available in Microsoft Fabric.
Conclusion
This provides a detailed insight into a Lakehouse in Microsoft Fabric, with details on what you need to know about Tables and Files in a Lakehouse and how to decide if a Lakehouse is the best for your Fabric Workload, as your use case.
About the Author
Musili Adebayo
Data Engineer | Fabric Analytics Engineer | Data Analyst | Azure | Power BI | Python | SQL
Adebayo, M (2025). Working with a Lakehouse in Microsoft Fabric - Everything you need to know. Available at: (6) Working with a Lakehouse in Microsoft Fabric - Everything you need to know. | LinkedIn [Accessed: 25th September 2025].
