hadoop-hdfs-issues mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Todd Lipcon (JIRA)" <j...@apache.org>
Subject [jira] [Commented] (HDFS-4953) enable HDFS local reads via mmap
Date Tue, 10 Sep 2013 19:33:57 GMT

    [ https://issues.apache.org/jira/browse/HDFS-4953?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13763385#comment-13763385
] 

Todd Lipcon commented on HDFS-4953:
-----------------------------------

bq. I envisioned these buffers just being a meg or two. (That's why I made the "huge buffers
scare me" comment). Anyone care to comment on whether this assumption was accurate?

I agree that this is best for performance. In benchmarks I've done of the existing SCR implementation,
I found that the max throughput was achieved for buffers of 1MB, 2MB, and 4MB, and any larger
started to see a performance hit of about 30-40% due to increased cache misses.
                
> 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
>             Fix For: HDFS-4949
>
>         Attachments: benchmark.png, HDFS-4953.001.patch, HDFS-4953.002.patch, HDFS-4953.003.patch,
HDFS-4953.004.patch, HDFS-4953.005.patch, HDFS-4953.006.patch, HDFS-4953.007.patch, HDFS-4953.008.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