Hi Lottiejanedev,
How are you doing today?, As per my understanding,
Yes, Liquid Clustering works best with directly stored columns, and it might not behave as expected when clustering on derived or generated fields. One possible issue is that generated fields are not physically stored in the table but instead computed at query time, which can interfere with how Liquid Clustering organizes data.
If you need to cluster on a derived field, a workaround is to materialize the generated field as a real column during ingestion rather than defining it as a computed column. You can do this by adding the transformation in your ETL process instead of relying on a generated column.
It’s also worth checking if your clustering column has high cardinality—Liquid Clustering works best when clustering on columns that are frequently filtered but don’t have too many unique values. If the derived field is too unique, the clustering might not be effective.
If you share more details about the specific generated field, I can suggest a better approach!
Regards,
Brahma
