Skip to main content
ReferenceApplies to: Unified engine deployments
Unified accepts Spark-style SQL with documented engine and connector restrictions. The same language is available through Workbench, SQL notebook cells, and live reference examples. Familiar syntax from another SQL system does not establish compatibility by itself.

Statement structure

Submit a single statement or use an example script that submits its statements in order. A semicolon terminates a statement. String literals, quoted identifiers, and comments can contain semicolons without separating statements.

Names and context

A session has a current catalog and namespace. An unqualified table uses that context; a fully qualified name names the destination explicitly. Table aliases shorten column references and resolve ambiguous names in joins. Function names and aliases are matched without ASCII case sensitivity. The session setting spark.sql.caseSensitive controls supported identifier matching.

Results and table changes

A SELECT returns rows. Metadata commands return descriptions or status. Writes and maintenance can compute or prepare output before their table commit is confirmed. Inspect the operation’s execution and write outcome rather than treating acceptance as completion.

Reference notation

Syntax forms in prose use square brackets for optional parts and descriptive names such as expression or table_name. Runnable SQL blocks use demo.tpcds for sample tables and demo.examples for the write lab. Replace these names with your catalog and namespace.
Last modified on November 4, 2026