[
https://issues.apache.org/jira/browse/HDFS-4953?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13707573#comment-13707573
]
Colin Patrick McCabe commented on HDFS-4953:
--------------------------------------------
well... zero copy for remote readers isn't possible, by definition :)
I cannae change the laws o' physics, cap'n.
We did some microbenchmarks where the existing short-circuit implementation
(SCR) got around 2500 MB/s, and zero copy reads (ZCR) got around 7500 MB/s. We
found that the advantage disappeared if we had to re-create the mmap on each
read through the file -- that's why this change includes a long-lived cache of
mmaps in ClientMmapManager.
> 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: 2.2.0
> Reporter: Colin Patrick McCabe
> Assignee: Colin Patrick McCabe
> Attachments: HDFS-4953.001.patch, HDFS-4953.002.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
