Explaining Databricks lakehouse to students: analogies that hold up

For a student info session, these one-liners survived follow-up questions: lakehouse = data lake (cheap storage, e.g. S3) plus warehouse (reliable SQL, e.g. Redshift) in one system with no copying; Delta Lake = Parquet files plus a transaction log (safe writes, time travel), sitting ON the storage, not the storage itself; SQL Warehouse = compute engine only, reading Delta in place (unlike classic Redshift, which keeps its own copy); Unity Catalog = 'the map and the lock' (catalog.schema.object names, permissions, lineage); MLflow = 'a lab notebook for models'; AI Functions = 'like UPPER(), but the function is an LLM'. Common confusions to pre-empt: 'Delta Lake is the S3 lake' and 'Databricks is just notebooks in the cloud'. With AWS as a sponsor, frame Redshift as 'two systems vs one', not as worse.