Use Liquid Clustering
When enabled, Delta Lake liquid clustering replaces traditional table partitioning and ZORDER for simplified data layout and optimized query performance.
When enabled, the table create scripts add ALTER TABLE ... CLUSTER BY (<columns>) using the table's key columns (with the effective-from date column included on history-bearing tables). On wide tables (more than 28 columns) the scripts also set 'delta.dataSkippingNumIndexedCols' = '-1' so data-skipping statistics cover the clustering columns. An object-level Databricks Cluster By clause in the object metadata replaces the generated clustering for that object.
When to Change This Setting
Keep the default N when the workspace does not support liquid clustering or the tables are small enough not to benefit. Enable it to let Databricks manage data layout on the key columns without hand-tuned partitioning.
Related Settings
| Setting | Interaction |
|---|---|
| Add Primary Keys | Declares the same key columns the clustering is built on |
| Table Properties | Companion table-level tuning applied at create time |
Notes
- This setting is part of the Databricks settings category.
- The default value for this setting is
N.