Microsoft Fabric Explained: OneLake, Lakehouse, and Unified Analytics

Stop wiring identities, networking, and copy jobs between Data Factory, Synapse, ADLS, and Power BI — Fabric collapses that stack into one tenant with one storage layer, but the consolidation comes with its own bill.

This walkthrough traces a full Fabric analytics architecture: source data lands via Data Factory into a Lakehouse on OneLake, where a Spark runtime and a serverless SQL endpoint read the same Delta Parquet files, and Power BI sits on top using Direct Lake to read straight from storage with no copy and no refresh schedule. The recurring design principle is open, shared storage with swappable compute — so the Lakehouse-vs-Warehouse choice comes down to whether your team lives in Spark or T-SQL, not where data sits. The main gotcha: capacity is a shared pool of compute units, so a runaway Spark job can throttle your Power BI reports, idle capacity still bills, and Direct Lake silently falls back to DirectQuery past its size guardrails.

For data engineers and architects deciding whether to migrate an existing Azure analytics platform onto Fabric — or just sizing a first capacity.

⏱️ Chapters:
0:00 Intro
0:04 The Problem Fabric Solves
0:35 The Reference Architecture
1:11 OneLake: The Storage Foundation
1:45 The Lakehouse and Medallion Flow
2:22 Lakehouse vs Warehouse
2:59 Direct Lake: The Power BI Payoff
3:37 Failure Modes and Trade-offs
4:17 Recap and Your Next Step

Subscribe for more architecture breakdowns that cover the trade-offs, not just the happy path.

Check the current Azure docs — cloud services change.

#MicrosoftFabric #OneLake #Lakehouse #DirectLake #AzureAnalytics

Видео Microsoft Fabric Explained: OneLake, Lakehouse, and Unified Analytics канала Joyjeet Majumdar