You are viewing a plain text version of this content. The canonical link for it is here.
Posted to common-issues@hadoop.apache.org by "Steve Loughran (JIRA)" <ji...@apache.org> on 2017/11/01 13:50:00 UTC
[jira] [Commented] (HADOOP-15001) AliyunOSS: provide one
memory-based buffering mechanism in outpustream to oss
[ https://issues.apache.org/jira/browse/HADOOP-15001?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16234085#comment-16234085 ]
Steve Loughran commented on HADOOP-15001:
-----------------------------------------
Lessons of S3A here (HADOOP-13208 & predecessors?)
* easy for code to generate data faster than you can upload: needs ability to block writer threads if the queue of pending uploads is too big. That's all writers using the same FS.
* off heap storage is potentially better, though unconstrained growth still an issue.
Look at {{S3ABlockOutputStream}} & helper classes
> AliyunOSS: provide one memory-based buffering mechanism in outpustream to oss
> -----------------------------------------------------------------------------
>
> Key: HADOOP-15001
> URL: https://issues.apache.org/jira/browse/HADOOP-15001
> Project: Hadoop Common
> Issue Type: Sub-task
> Components: fs/oss
> Affects Versions: 3.0.0-beta1
> Reporter: Genmao Yu
> Priority: Major
>
--
This message was sent by Atlassian JIRA
(v6.4.14#64029)
---------------------------------------------------------------------
To unsubscribe, e-mail: common-issues-unsubscribe@hadoop.apache.org
For additional commands, e-mail: common-issues-help@hadoop.apache.org