← Understanding of Databricks Data Intelligence Platform
11% domain · module 2
Catalog Explorer: catalogs, schemas, tables, views, lineage, Marketplace
Catalog Explorer is the analyst map
From Catalog Explorer you search assets, read comments, apply tags, preview data (if privileged), inspect lineage, and see owners and grants. The exam wants you fluent in what each object *is*.
Three-level namespace
Always think catalog.schema.object. Unqualified names depend on your default catalog/schema — a common exam trap.
- Catalog — top of a metastore (often one per environment:
prod,dev). - Schema (database) — subject area, e.g.
finance,gold. - Table / view / volume / function / model — the object.
Managed vs external tables
- Managed (UC managed) — Databricks owns the data files in a managed location.
DROP TABLEremoves data files (after a grace period / VACUUM semantics). - External — table metadata in UC, files live in an external location you registered.
DROP TABLEdrops the table definition; files remain in storage. That is the classic “SHOW TABLES is empty but the lake still has Parquet” scenario.
Views vs certified tables vs lineage
- View — saved query; no extra copy of data (unless materialized).
- Certified / trusted datasets — tagged or process-certified gold tables you should prefer for dashboards.
- Lineage — table/column graph across SQL, notebooks, jobs, dashboards. Lineage does not grant
SELECT.
Databricks Marketplace
Marketplace installs data products into your metastore. After install, objects are normal UC objects: you still need privileges. It is not a security bypass and not a SQL warehouse.