Weekly Machine Learning Roundup: OneLake Ingestion and Unified Analysis
This week's ML-adjacent Fabric updates focused on reducing two workflow frictions: getting local artifacts into OneLake, and moving between SQL, notebooks, and KQL analysis without re-learning each workload UI. Building on last week's “operationalize the platform” theme (safer ingestion, fewer embedded secrets, smoother Warehouse querying), these changes aim to reduce glue work once teams move beyond prototypes.
Microsoft Fabric: lower-friction ingestion and a more consistent analysis surface
OneLake File Explorer is now GA, addressing a common prototyping need: early datasets and artifacts often start on a developer machine (Excel, CSV, Parquet, images, intermediate outputs). With Windows File Explorer integration, OneLake mounts in Explorer so teams can browse by workspace/item and use standard file operations like drag-and-drop to place files where they belong. In the context of last week's Eventstreams ingestion and security improvements (private networking, Key Vault certs, fewer embedded connection strings), this is a complementary on-ramp: teams can move local artifacts into governed storage without scripts or portal detours. Once in OneLake, data is immediately usable across Fabric experiences (pipelines, notebooks, semantic models) without one-off uploads during iteration. In preview, Fabric is reducing UI fragmentation with a unified “Analyze data with” entry point across Lakehouse, Data Warehouse, and Eventhouse. This follows last week's “cleaner warehouse SQL” thread: once data is shared in OneLake, friction often shifts to inconsistent compute and query entry points. Eventhouse Endpoint now appears alongside SQL Endpoint and Notebook options so switching modalities is predictable from the same menu. For Lakehouse and Warehouse, enabling Eventhouse Endpoint provisions an Eventhouse and KQL Database as child artifacts with backend-managed schema sync, which provides a KQL surface over the same data without manual sync or duplication. That matches last week's push for managed configuration over bespoke integration. Eventhouse also gets the same menu at the database level (next to Share), and notebook launching is standardized so opening from Eventhouse/KQL Database auto-adds the database to the notebook environment for consistent Spark notebook behavior across workloads.