[
https://issues.apache.org/jira/browse/HADOOP-15124?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16523933#comment-16523933
]
Todd Lipcon commented on HADOOP-15124:
--------------------------------------
Before seeing this JIRA I also happened to have spent some time on this same
perf issue. My approach was just to micro-optimize the existing stats
implementation:
- use a simple array to iterate over the FS stats instead of iterating over a
HashSet (the latter involves a much more complex iterator)
- out-of-line the unlikely path for threadlocal (improve inlining)
- get rid of the visitor abstraction for visiting stats objects (it wasn't
getting escape-analyzed out or inlined, and was also causing actual boxing of
Longs)
In my teragen tests this also reduced the statistics to a small fraction of the
profile. I didn't compare vs a LongAdder approach, though.
My patch is at:
https://github.com/toddlipcon/hadoop-common/commit/e5bedddabbb9e8729b2f58165f0849c30e2be346
> Slow FileSystem.Statistics counters implementation
> --------------------------------------------------
>
> Key: HADOOP-15124
> URL: https://issues.apache.org/jira/browse/HADOOP-15124
> Project: Hadoop Common
> Issue Type: Sub-task
> Components: common
> Affects Versions: 2.9.0, 2.8.3, 2.7.5, 3.0.0, 3.1.0
> Reporter: Igor Dvorzhak
> Assignee: Igor Dvorzhak
> Priority: Major
> Labels: common, filesystem, fs, statistics
> Attachments: HADOOP-15124.001.patch
>
>
> While profiling 1TB TeraGen job on Hadoop 2.8.2 cluster (Google Dataproc, 2
> workers, GCS connector) I saw that FileSystem.Statistics code paths Wall time
> is 5.58% and CPU time is 26.5% of total execution time.
> After switching FileSystem.Statistics implementation to LongAdder, consumed
> Wall time decreased to 0.006% and CPU time to 0.104% of total execution time.
> Total job runtime decreased from 66 mins to 61 mins.
> These results are not conclusive, because I didn't benchmark multiple times
> to average results, but regardless of performance gains switching to
> LongAdder simplifies code and reduces its complexity.
--
This message was sent by Atlassian JIRA
(v7.6.3#76005)
---------------------------------------------------------------------
To unsubscribe, e-mail: common-issues-unsubscr...@hadoop.apache.org
For additional commands, e-mail: common-issues-h...@hadoop.apache.org
