Optimization Method Columns (Model Object)

Applies to
Platform: Databricks · Databricks dbt · Fabric Lakehouse

Optimization Method Columns (OptimizationMethodColumns) holds the terms whose columns the platform uses to partition or cluster a generated table. It works together with Optimization Method.

The property is not available on Microsoft SQL Server, Snowflake, Snowflake dbt, Fabric Warehouse and Oracle.

Where to set it

Set Optimization Method Columns when you edit a model object that has Optimization Method. The property is a term selector: you select one or more terms of the model object from its term list.

Values

One or more terms of the model object: used when Optimization Method is Partitioned By or Cluster By.
With Optimization Method None, the selected terms are ignored.
Empty while a method is set: validation reports the error Model object has an optimization method, but no optimization method columns defined. The client does not generate the project until a column is selected.

Default value

Empty, no term selected.

Effect on generated tables

Databricks and Fabric Lakehouse: the columns of the selected terms appear in PARTITIONED BY (...) or CLUSTER BY (...) of the table.
Databricks dbt: the term names appear in partition_by=[...] or liquid_clustered_by=[...] of the model configuration.

Example

The Hub load_customer_hub has Optimization Method Partitioned By and the business key term CUSTOMER_ID in Optimization Method Columns. On Databricks the deployment notebook creates the Hub table with:

USING delta
PARTITIONED BY (
     `customer_id`
)

In a Databricks dbt project the Hub model gets:

file_format='delta',
partition_by=['CUSTOMER_ID'],

A Satellite with Optimization Method Cluster By and the term DESCRIPTION gets CLUSTER BY (`description`) on Databricks and liquid_clustered_by=['DESCRIPTION'] in dbt.

Reference: Optimization Method (Model Object), File Format (Model Object)
See also: Optimize load performance