Yao Xu has posted comments on this change. ( http://gerrit.cloudera.org:8080/14046 )
Change subject: KUDU-2917 Split a tablet into primary key ranges by number of rows ...................................................................... Patch Set 6: (1 comment) > (1 comment) > > Can you quantify the performance impact using a benchmark? > > I'm also curious whether you looked into making the existing > chunk-by-size heuristic more accurate by considering encoded size, > factoring in deltas, or something along those lines. Sure, I'm going to implement a SplitKeyRange benchmark. I have considered adding a decode data size to the DiskRowSet to make splitKeyRange better. However, it is still difficult to resolve the case where column data is null. So I first implemented the feature of splitting key range by the number of rows. http://gerrit.cloudera.org:8080/#/c/14046/4/src/kudu/tablet/deltamemstore.cc File src/kudu/tablet/deltamemstore.cc: http://gerrit.cloudera.org:8080/#/c/14046/4/src/kudu/tablet/deltamemstore.cc@124 PS4, Line 124: : if (update.is_delete()) { : > The DMS doesn't allow REINSERTs. Maybe you can enforce that via DCHECK? You're right. I misremembered. :( -- To view, visit http://gerrit.cloudera.org:8080/14046 To unsubscribe, visit http://gerrit.cloudera.org:8080/settings Gerrit-Project: kudu Gerrit-Branch: master Gerrit-MessageType: comment Gerrit-Change-Id: Ib859c615743ebf7c3b481dfe1bd428769b02f0d7 Gerrit-Change-Number: 14046 Gerrit-PatchSet: 6 Gerrit-Owner: Yao Xu <[email protected]> Gerrit-Reviewer: Adar Dembo <[email protected]> Gerrit-Reviewer: Andrew Wong <[email protected]> Gerrit-Reviewer: Kudu Jenkins (120) Gerrit-Reviewer: Tidy Bot (241) Gerrit-Reviewer: Yao Xu <[email protected]> Gerrit-Comment-Date: Tue, 13 Aug 2019 09:47:39 +0000 Gerrit-HasComments: Yes
