Hi
@Lucas Zago -- check out the docs here --
https://docs.databricks.com/delta/data-skipping.html
You won't always want to z-order a table, only z-order when read performance is important (ie gold-level tables that will be queried by end users; source tables for streaming pipelines with low-latency requirements). Z-ordering will really help on columns with numeric data types, like primary and foreign keys in a star-schema data model. Z-ordering should not be done on every column, only columns that will most often be used in join conditions or predicates of queries.