Mailing-List: contact hdfs-issues-help@hadoop.apache.org; run by ezmlm
Precedence: bulk
Reply-To: hdfs-issues@hadoop.apache.org
Date: Mon, 12 Aug 2013 22:11:50 +0000 (UTC)
From: "Colin Patrick McCabe (JIRA)" <jira@apache.org>
To: hdfs-issues@hadoop.apache.org
Message-ID: <JIRA.12655953.1372813800093.49800.1376345510639@arcas>
In-Reply-To: <JIRA.12655953.1372813800093@arcas>
References: <JIRA.12655953.1372813800093@arcas>
Subject: [jira] [Commented] (HDFS-4953) enable HDFS local reads via mmap
MIME-Version: 1.0
Content-Type: text/plain; charset=utf-8
Content-Transfer-Encoding: 7bit


    [ https://issues.apache.org/jira/browse/HDFS-4953?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13737430#comment-13737430 ] 

Colin Patrick McCabe commented on HDFS-4953:
--------------------------------------------

HDFS-4949 is a related, but independent change.  In HDFS-4949, the DataNode manages regions of memory using {{mmap}} and {{mlock}}.  The {{mmap}} on the DataNode side is done only to make {{mlock}} possible.  The {{mmap}} on the client side is done to allow the client to read from the {{mmap}} rather than using the {[read}} syscall.  {{mmap}} regions are specific to a particular process-- an {{mmap}} in the DataNode is of no use to the client process.  ClientMmapManager is named the way it is to make it obvious that it's a *client* thing, not a DataNode thing.
                
> 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, HDFS-4953.006.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