If a table has many micro-partitions and a query scans all of them, what does that indicate about partition pruning?

• A. Partition pruning is not efficient
• B. Partition pruning is efficient
• C. No pruning is possible
• D. The table is not partitioned

Answer: (A)

Partition pruning works by skipping micro-partitions that cannot contain the query results based on the filters and the metadata Snowflake stores for those partitions. If a query ends up scanning all micro-partitions, pruning isn’t eliminating any partitions, so its effectiveness for that query is not good. This typically happens when there are no selective predicates on the partitioned columns or the filters are broad enough that nearly every partition could satisfy them. Snowflake’s zone maps and min/max metadata allow pruning only when the query predicates are selective with respect to the partitioning; otherwise you end up reading all partitions, which is less efficient.