DataYatriSpark, seen under the hood
Simulate
databricks/platform
beginner
Connecting…

SQL Warehouses: compute shaped for BI

Serverless vs Pro vs Classic, the caches behind a fast dashboard, and why size and cluster count are different dials.

Lesson 3 of 14 · Databricks path

Explain it at my level

  1. Endpoint
  2. Three types
  3. Caches
  4. Two dials
  5. Auto stop
Watch the canvas:BI client / cached resultwarehouse endpointcloud storageclassicpro / size dialserverless / scale outdisk cacheidle spendLive simulation
Client → endpoint → storageBI dashboardPower BI · Tableau · dbtSQL warehouseJDBC/ODBC · Photon · query queueCloud storageDelta tables · S3 / ADLSWarehouse typeClassicVMs in your cloud accountcold start 4–6 minProyour account · full SQLcold start 4–6 minServerlessDatabricks-managed warm poolcold start in secondsCache layersResult cacheidentical SQL · unchanged tables0 bytes scannedDisk cacheParquet on worker NVMe SSDno cloud round tripRemote readonly on a double missthe slow pathServer hostname + HTTP path — a warehouse is an endpoint, not a notebook.
One JDBC endpoint, three compute flavours, two caches, two scaling dials, and one off switch.