Metric view feature availability

Metric views were introduced in Databricks Runtime 16.4. The following table lists minimum Databricks Runtime version requirements for specific features added in later versions.

Note

You can also work with metric views on a SQL warehouse. A SQL warehouse is automatically updated with the latest Databricks SQL compute version, so you don't manage runtime versions the way you do for a cluster. See Channels.

Databricks Runtime version	Features
18.2	Parameters Wildcard expressions for fields and measures (YAML specification 1.1)
18.1	One-to-many joins (YAML specification 1.1) Window measure offset modifier (YAML specification 1.1) inclusive and exclusive modifiers on trailing and leading window ranges (YAML specification 1.1) Join optimization with rely.at_most_one_match
18.0	BI compatibility mode REFRESH MATERIALIZED VIEW
17.3	Snowflake schema Agent metadata: comments, synonyms, formatting, display names (YAML specification 1.1) TEMPORARY metric views Materialization JDBC/ODBC support through Thrift server

YAML specification versions

The metric view YAML specification is versioned independently of Databricks Runtime. The following sections describe what each version adds.

Version 1.1

Adds support for agent metadata (display names, formatting, synonyms) and the optional comment field for metric views, fields (also called dimensions), and measures. See Agent metadata in metric views.

Also adds the following fields and modifiers (require Databricks Runtime 18.1 or above):

• The cardinality field on joins, including the one_to_many value. See Join cardinality.
• The offset field on window specifications. See How offset shifts the window frame.
• The inclusive and exclusive modifiers on trailing and leading range values. See Include or exclude the anchor row.

Version 0.1

Initial release of the metric view YAML specification.