OneLake vs Lakehouse vs Warehouse vs Shortcut: which Fabric analytics piece fits

Verdict: Warehouse suits structured SQL data needing schema enforcement and concurrency; Lakehouse suits mixed or semi-structured data and Spark or ML workloads; both sit on OneLake, the single shared store; a shortcut references external data in place without copying it.

CriterionOneLakeLakehouseWarehouseShortcut
What it isSingle logical data lake built on ADLS Gen2, shared by every Fabric workloadFiles, folders, and tables acting as a database over the lakeManaged SQL Server compatible relational warehouse backed by OneLakeLive reference that makes external storage appear inside a Lakehouse with no copy
Use whenYou need one shared tenant-wide store underneath all Fabric itemsData is mixed or semi-structured and the team works in Spark notebooks or MLData is structured, the team writes SQL, and needs schema enforcement with high concurrencyData must stay in its original location, such as external S3 storage

Rules

Traps