Usecase is that as we work on https://github.com/apache/arrow-datafusion/issues/4085 we need to have some way to: 1. Verify page index filtering is actually working / enabled as expected 2. Evaluate its efficiency (how long it takes, etc) to see if further optimizations are needed What I would like is: 1. Add metrics for "how many rows were pruned using page level skipping" and "how long did the page level skipping code take" Related work is this PR that added similar metrics for filter pushdown (see https://github.com/apache/arrow-datafusion/pull/3989) -- we can probably follow a similar approach cc @Ted-Jiang
Activity
[-]Add statistics for parquet page level skipping[/-][+]Add metrics for parquet page level skipping[/+]