Microsoft Fabric: The Big Picture
What is Microsoft Fabric, why does it exist, and how does OneLake unify all your analytics data? Your foundation for everything in DP-600.
What is Microsoft Fabric?
Imagine a shopping mall for data.
Before Fabric, doing analytics was like visiting separate shops scattered across town — one shop for storing raw files, another for building data warehouses, a third for creating reports, and a fourth for running machine learning. Each shop had its own parking lot (storage), its own loyalty card (security), and its own opening hours (management).
Microsoft Fabric is the shopping mall that puts all these shops under one roof. One parking lot (OneLake), one loyalty card (Entra ID), one management office (admin portal). The shops still specialise — but they share everything underneath.
For the DP-600 exam, you need to understand the mall layout and know which shop to visit for each task.
The problem Fabric solves
Before Fabric, enterprise analytics looked like this:
| Challenge | Old World | Fabric World |
|---|---|---|
| Storage | Multiple storage accounts, duplicated data everywhere | OneLake — one copy of data, shared across all engines |
| Security | Different permissions per service, hard to audit | Entra ID everywhere, sensitivity labels, workspace RBAC |
| Data movement | Copy data between services (expensive, slow, error-prone) | Engines read directly from OneLake — no copies needed |
| Governance | Separate tools for lineage, classification, compliance | Microsoft Purview integration across all Fabric items |
| Billing | Per-service billing, hard to predict costs | Capacity-based — one pool of compute for everything |
Scenario: Meet our four analytics professionals
Throughout this course, you will follow four characters facing real Fabric challenges:
🛒 Anita Patel — Senior Analytics Engineer at FreshCart, a national grocery chain with 2,000+ stores. She builds sales dashboards, demand forecasting, and inventory analytics. Her data: daily POS transactions, supplier feeds, loyalty programme data.
🏢 James Okafor — BI Architect at Summit Consulting, a 400-person consulting firm managing analytics for 15+ clients. He designs multi-workspace environments with strict governance and reusable templates.
🏥 Dr. Sarah Nguyen — Health Data Analyst at Pacific Health Network, a 6,000-person hospital system. She builds patient outcome dashboards and clinical reporting while navigating strict compliance requirements.
💰 Raj Mehta — Fabric Platform Admin at Atlas Capital, a mid-size financial services firm. He manages Fabric capacity, monitors performance, and ensures every report meets audit requirements.
OneLake: One lake to rule them all
OneLake is the foundation of everything in Fabric. Every workspace, every lakehouse, every warehouse writes to OneLake.
Key OneLake concepts
- One per tenant — your entire organisation shares a single OneLake (like OneDrive is one drive per user, OneLake is one lake per tenant)
- Delta Parquet format — table data in lakehouses and warehouses is stored as Delta tables (Parquet files with a transaction log). OneLake can also hold non-Delta files.
- No data duplication — Spark, SQL, KQL, and Power BI all read the same physical files
- Shortcuts — virtual pointers to data in external storage (Azure Data Lake, S3, Google Cloud Storage) without copying it into OneLake
- Governed by default — Entra ID authentication, sensitivity labels, and Purview integration from day one
Exam tip: OneLake vs Azure Data Lake Storage
The exam may compare OneLake with Azure Data Lake Storage Gen2 (ADLS Gen2). Key differences: OneLake is tenant-scoped (one per org, not per subscription), uses Delta format by default (not raw Parquet/CSV), and is managed by Fabric (no storage account setup needed). ADLS Gen2 is still used for external data accessed via shortcuts.
Fabric workloads at a glance
Fabric bundles several workloads under one roof. Each workload has its own engine but shares OneLake storage.
| Workload | What It Does | Engine | Key Items |
|---|---|---|---|
| Data Engineering | Build and manage lakehouses, notebooks, Spark jobs | Apache Spark | Lakehouse, Notebook, Spark Job |
| Data Warehouse | SQL-based analytics warehouse with full T-SQL support | Fabric SQL Engine | Warehouse, SQL Endpoint |
| Data Factory | Data integration — pipelines and Dataflows Gen2 for ETL/ELT | Data movement engine | Pipeline, Dataflow Gen2 |
| Power BI | Semantic models, reports, dashboards for business users | Analysis Services | Semantic Model, Report, Dashboard |
| Real-Time Intelligence | Streaming analytics and time-series data | KQL (Kusto) | Eventhouse, KQL Queryset |
| Data Science | Machine learning experiments and models | Apache Spark + ML libraries | Experiment, ML Model |
Which workload for which task?
| If you need to… | Use this workload |
|---|---|
| Store raw files and build Delta tables | Data Engineering (Lakehouse) |
| Write complex SQL queries with joins and aggregations | Data Warehouse |
| Move data from external sources into Fabric | Data Factory (Pipeline or Dataflow Gen2) |
| Build interactive dashboards for business users | Power BI |
| Analyse streaming IoT or event data in near real-time | Real-Time Intelligence |
| Train and deploy ML models | Data Science |
Workspaces: Your unit of everything
Workspaces are how you organise Fabric items. Think of them as project folders with built-in security and collaboration.
- One workspace = one team or project — all related items live together
- Access control — workspace roles (Admin, Member, Contributor, Viewer) control who can do what
- Git integration — connect a workspace to a Git repo for version control
- Deployment pipelines — promote items from Dev → Test → Prod workspaces
- Capacity assignment — each workspace runs on an assigned Fabric capacity (F SKU)
Scenario: James sets up a client workspace
James at Summit Consulting creates a separate workspace for each client engagement. Client A’s workspace has its own lakehouse, warehouse, and semantic models — completely isolated from Client B. He assigns each workspace to the appropriate capacity and configures Git integration so his team can track changes.
This pattern — one workspace per client or project — is the foundation of multi-tenant governance in Fabric.
Fabric items you need to know
| Item | What It Is | Where It Lives |
|---|---|---|
| Lakehouse | A storage container with Delta tables and a SQL analytics endpoint | Data Engineering |
| Warehouse | A fully managed SQL data warehouse | Data Warehouse |
| Semantic Model | A Power BI data model that defines measures, relationships, and business logic | Power BI |
| Notebook | An interactive coding environment (PySpark, SQL, R) | Data Engineering / Data Science |
| Pipeline | An orchestration tool that moves and transforms data on a schedule | Data Factory |
| Dataflow Gen2 | A low-code/no-code data transformation tool (Power Query Online) | Data Factory |
| Eventhouse | A real-time analytics database for streaming data | Real-Time Intelligence |
| KQL Queryset | A Kusto Query Language query editor for Eventhouse data | Real-Time Intelligence |
Anita at FreshCart needs to store raw CSV files from 2,000 stores, transform them into structured tables, and build Power BI dashboards. She wants a single platform without managing separate Azure services. Which statement best describes what Microsoft Fabric offers?
James at Summit Consulting manages analytics for 15 clients. He needs to ensure each client's data is isolated with its own access controls. Which Fabric concept should he use as the primary isolation boundary?
Raj at Atlas Capital is evaluating Fabric. His CTO asks: 'What stops us from needing to copy financial data between services?' Which OneLake feature addresses this concern?
🎬 Video coming soon
Next up: Lakehouses: Your Data Foundation — the core storage item in Fabric, where your raw data becomes structured Delta tables.