Adar Dembo has posted comments on this change. ( 
http://gerrit.cloudera.org:8080/12196 )

Change subject: generic_iterators: prep for MergeIterator dominance
......................................................................


Patch Set 4:

(2 comments)

http://gerrit.cloudera.org:8080/#/c/12196/2//COMMIT_MSG
Commit Message:

http://gerrit.cloudera.org:8080/#/c/12196/2//COMMIT_MSG@9
PS2, Line 9: This patch adds a histogram to the MergeIterator that tracks the 
number of
> instead of a histogram isn't what we really care about just the total numbe
I originally went with a histogram because in addition to the total number of 
comparisons, I also wanted to get a sense for how they were distributed across 
the lifetime of the MergeIterator. Spitting out the percentiles as I did 
doesn't really accomplish that, nor does the distribution really matter for the 
purposes of computing the increase in efficiency (we assume that every scan 
runs to completion, so the total CPU consumption is the only thing that really 
matters).

Anyway, I'll switch this to a raw counter.


http://gerrit.cloudera.org:8080/#/c/12196/1/src/kudu/common/generic_iterators.h
File src/kudu/common/generic_iterators.h:

http://gerrit.cloudera.org:8080/#/c/12196/1/src/kudu/common/generic_iterators.h@184
PS1, Line 184:   HdrHistogram sub_iter_histo_;
> I see the appeal of this for your testing, but does this have adverse perf
I don't know; I didn't measure the impact of the histogram itself. Looking at 
the implementation, it's conceivable that it may have an impact, as it does 
some arithmetic, atomic increments, and CASes.

Are you open to a counter, even if it was only used in tests? IteratorStats 
would probably be the best home, but those are maintained on a per-column 
basis; "number of key comparisons" doesn't really make sense in that context, 
unless you want it broken up by key column. And beyond that, it's a metric that 
very clearly only matters if using ORDERED scans.



--
To view, visit http://gerrit.cloudera.org:8080/12196
To unsubscribe, visit http://gerrit.cloudera.org:8080/settings

Gerrit-Project: kudu
Gerrit-Branch: master
Gerrit-MessageType: comment
Gerrit-Change-Id: I3a8db5973dc09b7271e05b3cc28025b7a2a9e21b
Gerrit-Change-Number: 12196
Gerrit-PatchSet: 4
Gerrit-Owner: Adar Dembo <a...@cloudera.com>
Gerrit-Reviewer: Adar Dembo <a...@cloudera.com>
Gerrit-Reviewer: Grant Henke <granthe...@apache.org>
Gerrit-Reviewer: Kudu Jenkins (120)
Gerrit-Reviewer: Mike Percy <mpe...@apache.org>
Gerrit-Reviewer: Todd Lipcon <t...@apache.org>
Gerrit-Comment-Date: Fri, 11 Jan 2019 23:53:50 +0000
Gerrit-HasComments: Yes

Reply via email to