Lars Volker has posted comments on this change. ( http://gerrit.cloudera.org:8080/12065 )
Change subject: IMPALA-5843: Use page index in Parquet files to skip pages ...................................................................... Patch Set 9: > (54 comments) > > In a few comments the row group level profile counters came up. > Whether we should only have "NumStatsFilteredRowGroups" and count > row groups that are filtered out by either row group-level stats or > page-level stats. > Or, we should have multiple variants of that counter, counting the > row group-level and page-level filtering separately. What do you > think? For debugging it's good to have more fine-grained data, but > I'm not sure that the query profile is the right place for such > information. I think having counter for row groups filtered based on row group stats, and pages filtered based on page stats would be good (and sufficient). I think entire row groups filtered by page stats (but not row group stats) are not worth tracking in a separate counter. The distinction I'd make is whether we discarded a row group without looking at the index or not. -- To view, visit http://gerrit.cloudera.org:8080/12065 To unsubscribe, visit http://gerrit.cloudera.org:8080/settings Gerrit-Project: Impala-ASF Gerrit-Branch: master Gerrit-MessageType: comment Gerrit-Change-Id: I0cc99f129f2048dbafbe7f5a51d1ea3a5005731a Gerrit-Change-Number: 12065 Gerrit-PatchSet: 9 Gerrit-Owner: Zoltan Borok-Nagy <[email protected]> Gerrit-Reviewer: Csaba Ringhofer <[email protected]> Gerrit-Reviewer: Impala Public Jenkins <[email protected]> Gerrit-Reviewer: Lars Volker <[email protected]> Gerrit-Reviewer: Michael Ho <[email protected]> Gerrit-Reviewer: Pooja Nilangekar <[email protected]> Gerrit-Reviewer: Tim Armstrong <[email protected]> Gerrit-Reviewer: Zoltan Borok-Nagy <[email protected]> Gerrit-Comment-Date: Tue, 09 Apr 2019 21:55:19 +0000 Gerrit-HasComments: No
