Microsoft Fabric Cheat Sheet

Updated 2026-04-29

Next Topic: Modern PaaS Platforms Cheat Sheet

🧠Study flashcards on this topic101 cards · spaced repetition→

Microsoft Fabric is a unified, end-to-end analytics platform that combines data integration, engineering, warehousing, science, real-time intelligence, and business intelligence into a single SaaS offering. Built on OneLake — a centralized data lake storing all data in open Delta/Parquet format — Fabric eliminates data silos and enables seamless collaboration across teams. The platform uses a capacity-based pricing model measured in Compute Units (CUs) purchased as F-SKUs, where workloads draw from a shared pool. As of 2026, Fabric serves over 31,000 customers and has introduced major additions including SQL Database in Fabric, Materialized Lake Views, Fabric IQ, OneLake Security, and an MCP Server integration — understanding these alongside the foundational item types is essential for building scalable, governed data solutions.

What This Cheat Sheet Covers

This topic spans 15 focused tables and 115 indexed concepts, 101 flashcards. Below is a complete table-by-table outline of this topic, spanning foundational concepts through advanced details.

Table 1: Core Platform ComponentsTable 2: Data Storage & IntegrationTable 3: Data Engineering WorkloadsTable 4: Data Pipelines & OrchestrationTable 5: Real-Time IntelligenceTable 6: Power BI IntegrationTable 7: Data Science & AITable 8: Governance & SecurityTable 9: Capacity ManagementTable 10: Development & CI/CDTable 11: Item Types ReferenceTable 12: Pricing & LicensingTable 13: Advanced FeaturesTable 14: Query Languages & ToolsTable 15: Best Practices & Patterns

Quick IndexSubscribe to unlock

A jump-to index of every table row in this cheat sheet.

Mind MapSubscribe to unlock

An interactive map of every table and concept in this topic.

Table 1: Core Platform Components

These are the structural pieces every Fabric solution is built from — the lake that holds the data, the workspaces and capacity that organize and power it, and the item types like lakehouse, warehouse, semantic model, and eventhouse where the actual work happens. Grasp how these nest inside one another and you understand most of how Fabric is laid out, right up to the newer governance and AI layers such as the OneLake Catalog, Fabric IQ, and Domains.

Component	Example	Description
OneLake	All workspaces store data in `OneLake/MyWorkspace/MyLakehouse/`	• Single unified data lake for all Fabric data • automatically provisioned with every tenant • stores data in open Delta Parquet format • supports ADLS Gen2 APIs.
Workspace	`Sales Analytics Workspace`	• Logical container for Fabric items (lakehouses, pipelines, reports) • assigned to a capacity • supports role-based access (Admin, Member, Contributor, Viewer).
Capacity	F64 SKU = 64 CUs	• Compute and storage pool purchased as SKUs (F2–F2048) • measured in Capacity Units • supports pay-as-you-go or reserved pricing • workloads consume CUs dynamically.
Lakehouse	Bronze/Silver/Gold lakehouse layers	• Combines data lake flexibility with warehouse querying • stores structured/unstructured data in Delta format • includes automatic SQL analytics endpoint.
Warehouse	Enterprise data warehouse	• Fully managed T-SQL engine for structured analytics • optimized for BI reporting • supports stored procedures, views, security.
Semantic Model	Sales semantic model with DAX measures	• Business logic layer for Power BI reports • defines metrics, relationships, hierarchies • supports Direct Lake, Import, or DirectQuery modes.
Eventhouse	Real-time telemetry store	• Container for KQL databases • optimized for time-series and streaming data • supports millisecond-latency queries with Kusto Query Language.
OneLake Catalog	Search for `sales` across all workspaces	• Centralized discovery and governance hub for all Fabric items • three tabs: Explore, Govern, Secure • embedded in Teams, Excel, Copilot Studio • supports programmatic search via REST API.
Fabric IQ	Business ontology maps `Revenue` to semantic model measures	• Semantic intelligence layer that unifies analytical and operational data in a shared ontology • exposes business entities, relationships, and rules to AI agents • powers Planning in Fabric IQ and operations agents.
Domain	Marketing, Finance, HR domains	• Organizational unit for data mesh architecture • groups workspaces by business area • enables decentralized ownership and federated governance.

Table 2: Data Storage & Integration

The promise of OneLake is that data lives in one open Delta/Parquet store and you bring it in without copying it everywhere. These concepts are how you do that — shortcuts that reference external data in place, mirroring that replicates databases in near real time, and the security and SQL-endpoint layers that govern and expose it once it lands.

Concept	Example	Description
OneLake Shortcuts	Shortcut to ADLS Gen2 blob container	• Virtual references to external data (ADLS Gen2, S3, GCS, other OneLake locations) • no data movement • supports Delta, Parquet, CSV • appears as native Fabric data.

The rest of this section is part of the full cheat sheet

Table 3: Data Engineering Workloads

Spark is the engine that powers heavy data transformation in Fabric, and these items are the surfaces you use to drive it — interactive notebooks, packaged job definitions, and the auto-refreshing Materialized Lake Views that let you build medallion pipelines declaratively. The remaining rows cover what shapes performance and cost: the runtime version, reusable environments, high-concurrency sharing of a single Spark session, and pre-warmed pools that kill cold-start delays.

Item	Example	Description
Notebook	PySpark data transformation `df = spark.read.format("delta").load("...")`	• Interactive coding environment for Spark (Python, Scala, R, SQL) • supports auto-save, Git integration, parameterization • edit directly in VS Code with live workspace sync.

The rest of this section is part of the full cheat sheet

Table 4: Data Pipelines & Orchestration

Carried over from Azure Data Factory, this is Fabric's ETL/ELT layer — how you move data in and chain steps together on a schedule. Note the split between full pipelines (orchestrating hundreds of activities) and the lighter Copy Job, which handles incremental and CDC movement without you authoring a pipeline at all; Dataflow Gen2 covers the low-code Power Query transformations in between.

Activity	Example	Description
Data Pipeline	Copy → Notebook → Stored Proc	• Orchestration workflow for ETL/ELT • supports 400+ activities (Copy Data, Dataflow, Notebook, Web, Script) • Azure Data Factory heritage.

The rest of this section is part of the full cheat sheet

Table 1: Core Platform Components

Component	Example	Description
OneLake	All workspaces store data in `OneLake/MyWorkspace/MyLakehouse/`	• Single unified data lake for all Fabric data • automatically provisioned with every tenant • stores data in open Delta Parquet format • supports ADLS Gen2 APIs.
Workspace	`Sales Analytics Workspace`	• Logical container for Fabric items (lakehouses, pipelines, reports) • assigned to a capacity • supports role-based access (Admin, Member, Contributor, Viewer).
Capacity	F64 SKU = 64 CUs	• Compute and storage pool purchased as SKUs (F2–F2048) • measured in Capacity Units • supports pay-as-you-go or reserved pricing • workloads consume CUs dynamically.
Lakehouse	Bronze/Silver/Gold lakehouse layers	• Combines data lake flexibility with warehouse querying • stores structured/unstructured data in Delta format • includes automatic SQL analytics endpoint.
Warehouse	Enterprise data warehouse	• Fully managed T-SQL engine for structured analytics • optimized for BI reporting • supports stored procedures, views, security.
Semantic Model	Sales semantic model with DAX measures	• Business logic layer for Power BI reports • defines metrics, relationships, hierarchies • supports Direct Lake, Import, or DirectQuery modes.
Eventhouse	Real-time telemetry store	• Container for KQL databases • optimized for time-series and streaming data • supports millisecond-latency queries with Kusto Query Language.
OneLake Catalog	Search for `sales` across all workspaces	• Centralized discovery and governance hub for all Fabric items • three tabs: Explore, Govern, Secure • embedded in Teams, Excel, Copilot Studio • supports programmatic search via REST API.
Fabric IQ	Business ontology maps `Revenue` to semantic model measures	• Semantic intelligence layer that unifies analytical and operational data in a shared ontology • exposes business entities, relationships, and rules to AI agents • powers Planning in Fabric IQ and operations agents.
Domain	Marketing, Finance, HR domains	• Organizational unit for data mesh architecture • groups workspaces by business area • enables decentralized ownership and federated governance.

Table 2: Data Storage & Integration

Concept	Example	Description
OneLake Shortcuts	Shortcut to ADLS Gen2 blob container	• Virtual references to external data (ADLS Gen2, S3, GCS, other OneLake locations) • no data movement • supports Delta, Parquet, CSV • appears as native Fabric data.

Table 3: Data Engineering Workloads

Item	Example	Description
Notebook	PySpark data transformation `df = spark.read.format("delta").load("...")`	• Interactive coding environment for Spark (Python, Scala, R, SQL) • supports auto-save, Git integration, parameterization • edit directly in VS Code with live workspace sync.

Table 4: Data Pipelines & Orchestration

Activity	Example	Description
Data Pipeline	Copy → Notebook → Stored Proc	• Orchestration workflow for ETL/ELT • supports 400+ activities (Copy Data, Dataflow, Notebook, Web, Script) • Azure Data Factory heritage.