OPTIMIZE [db_name.]table_name [WHERE predicate] [ZORDER BY (col_name1, col_name2, ...)]
Optimize the layout of Delta Lake data. Optionally optimize a subset of data or colocate data by column. If you do not specify colocation, bin-packing optimization is performed.
Bin-packing optimization is idempotent, meaning that if it is run twice on the same dataset, the second run has no effect. It aims to produce evenly-balanced data files with respect to their size on disk, but not necessarily number of tuples per file. However, the two measures are most often correlated.
Z-Ordering is not idempotent but aims to be an incremental operation. The time it takes for Z-Ordering is not guaranteed to reduce over multiple runs. However, if no new data was added to a partition that was just Z-Ordered, another Z-Ordering of that partition will not have any effect. It aims to produce evenly-balanced data files with respect to the number of tuples, but not necessarily data size on disk. The two measures are most often correlated, but there can be situations when that is not the case, leading to skew in optimize task times.
To control the output file size, set the Spark configuration
spark.databricks.delta.optimize.maxFileSize. The default value is
1073741824, which sets the size to 1 GB. Specifying the value
104857600sets the file size to 100 MB.
Optimize the subset of rows matching the given partition predicate. Only filters involving partition key attributes are supported.
Colocate column information in the same set of files. Co-locality is used by Delta Lake data-skipping algorithms to dramatically reduce the amount of data that needs to be read. You can specify multiple columns for
ZORDER BY as a comma-separated list. However, the effectiveness of the locality drops with each additional column.