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 2016/12/01 15:34:58 UTC
[jira] [Commented] (HADOOP-13853) S3ADataBlocks.DiskBlock to lazy
create dest file for faster 0-byte puts
[ https://issues.apache.org/jira/browse/HADOOP-13853?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15712269#comment-15712269 ]
Steve Loughran commented on HADOOP-13853:
-----------------------------------------
note that given all the other overheads of a commit, this is not the bottleneck.
> S3ADataBlocks.DiskBlock to lazy create dest file for faster 0-byte puts
> -----------------------------------------------------------------------
>
> Key: HADOOP-13853
> URL: https://issues.apache.org/jira/browse/HADOOP-13853
> Project: Hadoop Common
> Issue Type: Sub-task
> Components: fs/s3
> Affects Versions: 2.8.0
> Reporter: Steve Loughran
> Priority: Minor
>
> Looking at traces of work, there's invariably a PUT of a _SUCCESS at the end, which, with disk output, adds the overhead of creating, writing to and then reading a 0 byte file.
> With a lazy create, the creation could be postponed until the first write, with special handling in the {{startUpload()}} operation to return a null stream, rather than reopen the file. Saves on some disk IO: create, read, delete
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
---------------------------------------------------------------------
To unsubscribe, e-mail: common-issues-unsubscribe@hadoop.apache.org
For additional commands, e-mail: common-issues-help@hadoop.apache.org