You are viewing a plain text version of this content. The canonical link for it is here.
Posted to issues@spark.apache.org by "Li Yuanjian (JIRA)" <ji...@apache.org> on 2017/12/13 12:31:01 UTC
[jira] [Updated] (SPARK-2926) Add MR-style (merge-sort)
SortShuffleReader for sort-based shuffle
[ https://issues.apache.org/jira/browse/SPARK-2926?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ]
Li Yuanjian updated SPARK-2926:
-------------------------------
Attachment: Spark Shuffle Test Report on Spark2.x.pdf
[~jerryshao] Hi saisai, thanks for your advise, I added a test report according to your suggestion. As described in the report, I only compare two shuffle mode in 'sort-by-key' workload because other test workloads shared same code paths in POC implementation(SortShuffleWriter with BlockStoreShuffleReader).
Also add a config( [code link|https://github.com/apache/spark/pull/19745/commits/fe9394eadf8ea51af2b2cb41b5b42981fa600752] ) just to force shutting down SerializedShuffle in 'sort-by-key' workload, otherwise both of master and POC use the SerializedShuffle.
For sort-by-key work around after closing Serialized Shuffle, the POC version can brings 1.44x faster than current master, although map side stage 1.16x slower, but reducer stage has 9.4x boosting.
> Add MR-style (merge-sort) SortShuffleReader for sort-based shuffle
> ------------------------------------------------------------------
>
> Key: SPARK-2926
> URL: https://issues.apache.org/jira/browse/SPARK-2926
> Project: Spark
> Issue Type: Improvement
> Components: Shuffle
> Affects Versions: 1.1.0
> Reporter: Saisai Shao
> Assignee: Saisai Shao
> Attachments: SortBasedShuffleRead.pdf, SortBasedShuffleReader on Spark 2.x.pdf, Spark Shuffle Test Report on Spark2.x.pdf, Spark Shuffle Test Report(contd).pdf, Spark Shuffle Test Report.pdf
>
>
> Currently Spark has already integrated sort-based shuffle write, which greatly improve the IO performance and reduce the memory consumption when reducer number is very large. But for the reducer side, it still adopts the implementation of hash-based shuffle reader, which neglects the ordering attributes of map output data in some situations.
> Here we propose a MR style sort-merge like shuffle reader for sort-based shuffle to better improve the performance of sort-based shuffle.
> Working in progress code and performance test report will be posted later when some unit test bugs are fixed.
> Any comments would be greatly appreciated.
> Thanks a lot.
--
This message was sent by Atlassian JIRA
(v6.4.14#64029)
---------------------------------------------------------------------
To unsubscribe, e-mail: issues-unsubscribe@spark.apache.org
For additional commands, e-mail: issues-help@spark.apache.org