You are viewing a plain text version of this content. The canonical link for it is here.
Posted to issues@spark.apache.org by "He Tianyi (JIRA)" <ji...@apache.org> on 2016/03/28 04:48:25 UTC
[jira] [Created] (SPARK-14186) Size awareness during spill and
shuffle
He Tianyi created SPARK-14186:
---------------------------------
Summary: Size awareness during spill and shuffle
Key: SPARK-14186
URL: https://issues.apache.org/jira/browse/SPARK-14186
Project: Spark
Issue Type: Improvement
Components: Shuffle, Spark Core
Affects Versions: 1.6.0
Reporter: He Tianyi
Priority: Minor
Currently during spilling of a collection of record, sorter calls {{createTempShuffleBlock}} for allocating a local block.
This call provides no size information about required block. Without knowledge of estimated size, this makes customized allocation strategy hard to implement .
On the other hand, theoretically both {{PartitionedPairBuffer}} and {{PartitionedAppendOnlyMap}} are able to track its estimated size.
In this case, it is convenient to extract size information and pass it as parameter to {{createTempShuffleBlock}}.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
---------------------------------------------------------------------
To unsubscribe, e-mail: issues-unsubscribe@spark.apache.org
For additional commands, e-mail: issues-help@spark.apache.org