Understand column usage & clustering health

The tables page now features a brand new columns tab, designed to help you understand:

1. How often each column in a table is being accessed
2. The clustering health of each column in your table. A score of 100% means the column is well clustered, and therefore including a filter on this column in your query will likely result in very good query pruning. We’ve found this feature incredibly helpful on our internal Snowflake account, as we’ve learned about columns we previously didn’t know were well clustered (i.e. query_id in this screenshot). This feature is also very helpful for understanding how tables without automatic clustering enabled are naturally clustered.

In conjunction with these, you can now filter workloads by the columns they’ve accessed. In the example below, you can see that all columns accessing the query_id column are pruning very well. The pruning efficiencies of over 90% indicate that most micro-partitions are being automatically removed.