OneLake comes with every Microsoft Fabric 6 tenant, exactly one per tenant: a data lake on Azure 6 Data Lake Storage, organized by workspaces and reachable through the ADLS Gen2 APIs. Every Fabric engine (T-SQL warehouse, Spark 129 , the Power BI engine through Direct Lake) stores its tables there as Delta Parquet 129 , so one copy serves all of them. Metadata virtualization makes the formats interchangeable: Iceberg 129 tables written to OneLake or reached through a shortcut appear as Delta to Fabric, and Delta tables appear as Iceberg to outside readers such as Snowflake, much as UniForm does for Databricks 2,717 (UniForm).
Two features have no counterpart in this chapter's open stack. Shortcuts mount S3, S3-compatible stores, ADLS, Google Cloud 1 Storage and Iceberg sources into OneLake without copying, and mirroring replicates operational databases into Delta tables continuously. OneLake security roles restrict folders, tables, rows and columns once and enforce them in SQL, Spark and Power BI alike.
Storage is billed per GB-month (East US): $0.026 hot, $0.019 cool and $0.004 cold. Everything else, including the compute behind OneLake reads and writes, consumes the workspace's capacity at $0.18 per capacity-unit hour pay-as-you-go (Synapse and Fabric), shared by all workloads and paused to stop billing. Fabric suits organizations already on Power BI and Microsoft 365, where Direct Lake reports over Delta tables replace scheduled imports.