You are viewing a plain text version of this content. The canonical link for it is here.
Posted to common-dev@hadoop.apache.org by "Christian Kunz (JIRA)" <ji...@apache.org> on 2009/01/22 02:33:59 UTC
[jira] Commented: (HADOOP-3672) support for persistent connections
to improve random read performance.
[ https://issues.apache.org/jira/browse/HADOOP-3672?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=12666028#action_12666028 ]
Christian Kunz commented on HADOOP-3672:
----------------------------------------
Unfortunately, this issue ended with mere discussion.
I would like to revive it to get it done, and I wonder whether Raghu's comment of 14/Jul/08 02:43 PM could be good starting point:
I think a very simple connection sharing at DFSClient.BlockReader() level for preads will do the job.
> support for persistent connections to improve random read performance.
> ----------------------------------------------------------------------
>
> Key: HADOOP-3672
> URL: https://issues.apache.org/jira/browse/HADOOP-3672
> Project: Hadoop Core
> Issue Type: Improvement
> Components: dfs
> Affects Versions: 0.17.0
> Environment: Linux 2.6.9-55 , Dual Core Opteron 280 2.4Ghz , 4GB memory
> Reporter: George Wu
> Attachments: pread_test.java
>
>
> preads() establish new connections per request. yourkit java profiles show that this connection overhead is pretty significant on the DataNode.
> I wrote a simple microbenchmark program which does many iterations of pread() from different offsets of a large file. I hacked DFSClient/DataNode code to re-use the same connection/DataNode request handler thread. The performance improvement was 7% when the data is served from disk and 80% when the data is served from the OS page cache.
--
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.