OneLake explained: the unified data lake of Microsoft Fabric

João Barros 19 de July de 2024 2 min read

OneLake is the central data lake of Microsoft Fabric. Unlike previous solutions (ADLS, Blob Storage, etc.), OneLake is a single repository for the whole organization — each Microsoft 365 tenant has exactly one OneLake, divided into workspaces.

Architecture

Tenant
└── OneLake (one per tenant)
    ├── Workspace: Analytics
    │   ├── Lakehouse: LH_Bronze
    │   ├── Lakehouse: LH_Silver
    │   └── Warehouse: WH_Prod
    └── Workspace: Marketing
        └── Lakehouse: LH_Campaigns

Data is stored in Delta Parquet format by default, ensuring compatibility with Spark, SQL and Power BI Direct Lake.

Shortcuts: access without moving data

A Shortcut is like a symbolic link — it points to data in another location (another Lakehouse, ADLS Gen2, S3, GCS) without copying it. Different teams can access the same data without duplication.

-- Create a shortcut via the Fabric API
POST https://api.fabric.microsoft.com/v1/workspaces/{workspaceId}/lakehouses/{lakehouseId}/shortcuts
{
  "path": "Tables/FactSales",
  "name": "FactSales_Shortcut",
  "target": {
    "type": "AdlsGen2",
    "location": "https://myadls.dfs.core.windows.net/container/factsales"
  }
}

Open Mirroring

Fabric supports mirroring of external databases (Azure SQL, Cosmos DB, Snowflake) directly into OneLake, with near-real-time replication and no need for complex ETL pipelines.

Conclusion

OneLake eliminates the proliferation of data silos. By centralizing all data in a single platform with unified access, organizations reduce complexity, integration costs and latency between data teams.