hadoop-hdfs-issues mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Tsuyoshi OZAWA (JIRA)" <j...@apache.org>
Subject [jira] [Commented] (HDFS-4953) enable HDFS local reads via mmap
Date Fri, 26 Jul 2013 04:11:51 GMT

    [ https://issues.apache.org/jira/browse/HDFS-4953?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13720384#comment-13720384
] 

Tsuyoshi OZAWA commented on HDFS-4953:
--------------------------------------

This work is very interesting to increase the throughput of HDFS. IMHO, more benchmarks are
needed to clarify when to use this feature. For instance:

1. Read large objects which doesn't fit in memory.
2. Basic benchmarks on CentOS 5. Old linux kernels cause page faults with the giant lock when
user program touches virtual memory which is mapped to a file on disk by mmap(2). This can
be problem for CDH users, because CDH supports CentOS 5[1].

[1] http://lwn.net/Articles/200215/
                
> enable HDFS local reads via mmap
> --------------------------------
>
>                 Key: HDFS-4953
>                 URL: https://issues.apache.org/jira/browse/HDFS-4953
>             Project: Hadoop HDFS
>          Issue Type: New Feature
>    Affects Versions: 2.3.0
>            Reporter: Colin Patrick McCabe
>            Assignee: Colin Patrick McCabe
>         Attachments: benchmark.png, HDFS-4953.001.patch, HDFS-4953.002.patch, HDFS-4953.003.patch,
HDFS-4953.004.patch, HDFS-4953.005.patch
>
>
> Currently, the short-circuit local read pathway allows HDFS clients to access files directly
without going through the DataNode.  However, all of these reads involve a copy at the operating
system level, since they rely on the read() / pread() / etc family of kernel interfaces.
> We would like to enable HDFS to read local files via mmap.  This would enable truly zero-copy
reads.
> In the initial implementation, zero-copy reads will only be performed when checksums
were disabled.  Later, we can use the DataNode's cache awareness to only perform zero-copy
reads when we know that checksum has already been verified.

--
This message is automatically generated by JIRA.
If you think it was sent incorrectly, please contact your JIRA administrators
For more information on JIRA, see: http://www.atlassian.com/software/jira

Mime
View raw message