Start here

Official Microsoft documentation tells you what every Fabric feature is. fabdocs.dev is about how to run it in production — how to build, deploy, govern, and scale lakehouses, notebooks, and pipelines without torching your capacity budget.

Who this is for

Data engineers taming lakehouses, Spark jobs, and Delta table sprawl.
Enterprise architects designing OneLake topology, security, and CI/CD.
Analytics leads who own a capacity and need CU burn to be predictable.

Where to start

New to Fabric's storage model?

Start with OneLake & Delta Lake — every other section assumes you already understand it.

Evaluating or migrating from another platform?

Honest comparisons first, then a step-by-step playbook once you've decided.

Own a capacity and need predictable CU burn?

Model Spark vs. SQL warehouse cost before you buy a bigger SKU.

Setting up governance or CI/CD from scratch?

OneLake security, workspace identity, and Git-backed deployment pipelines.

How the manual is organized

OneLake & Delta Lake

Delta optimization, V-Order, vacuum & retention, multi-cloud shortcuts.

Data Engineering

Production PySpark patterns, the Native Execution Engine, resource profiles, Runtime upgrades.

CI/CD & Git

Git integration, deployment pipelines, variable libraries, GitHub Actions for Fabric.

Governance & Security

OneLake security, workspace identity, audit logging for compliance.

Tools & Calculators

CU cost calculator, VS Code sync, capacity simulation.

Migration Playbooks

Step-by-step conversions from SSIS, Snowflake, and Databricks.

Conventions used in this manual

Callout	Meaning
CU impact	How the pattern changes Capacity Unit consumption.
Runtime	Minimum Fabric Runtime / Spark version required.
Reversible?	Whether the change can be rolled back without a reload.

Pages about compute-heavy operations — Delta maintenance, Spark tuning, capacity sizing — state CU impact up front, as the headline, not a footnote. Governance and CI/CD pages don't force it in where it isn't the relevant axis.