Creating tables, views, and materialized views in the catalog

In Azure Databricks, you use standard SQL DDL to create managed or external tables (Delta by default), logical views that re-run their defining query on each access, and materialized views that precompute and persist query results for faster reads. Materialized views are created and managed through Databricks SQL or Lakeflow/Delta Live Tables pipelines and require Unity Catalog.

1 · Learn the must-know

• 1CREATE TABLE without a LOCATION clause creates a managed Delta table whose data and metadata lifecycle is fully controlled by Unity Catalog; adding LOCATION creates an external table referencing existing storage.
• 2CREATE VIEW defines a virtual table stored only as a saved query (no data materialization); its results always reflect the current state of underlying tables at query time.
• 3CREATE MATERIALIZED VIEW precomputes and stores results in Delta format, giving faster query performance for expensive aggregations or joins, but the stored results only reflect data as of the last refresh.
• 4Materialized views require Unity Catalog and are refreshed automatically on a schedule or incrementally as source data changes, similar to how streaming tables are maintained.
• 5You cannot directly INSERT, UPDATE, or DELETE rows in a materialized view; changes must flow through its defining query and refresh process.
• 6Both views and materialized views can reference multiple tables or other views, but circular references are not allowed and materialized views cannot be created from streaming sources without appropriate incremental refresh support.

2 · Check your understanding