giraph-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Claudio Martella (JIRA)" <j...@apache.org>
Subject [jira] [Commented] (GIRAPH-514) DiskBackedMessageStores should take advantage of machines with multiple disks
Date Thu, 14 Feb 2013 23:11:12 GMT

    [ https://issues.apache.org/jira/browse/GIRAPH-514?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13578795#comment-13578795
] 

Claudio Martella commented on GIRAPH-514:
-----------------------------------------

yeah, it was just a test to see how bad it would was. 
Yes, I meant a flush over the network for edges. So yes, it was not connected with the additional
thread... Still, I cannot understand what was wrong, The job failed as the workers lost reservations
for partitions in inputsplits. Will try to replicate this tomorrow.
                
> DiskBackedMessageStores should take advantage of machines with multiple disks
> -----------------------------------------------------------------------------
>
>                 Key: GIRAPH-514
>                 URL: https://issues.apache.org/jira/browse/GIRAPH-514
>             Project: Giraph
>          Issue Type: Improvement
>            Reporter: Claudio Martella
>            Assignee: Claudio Martella
>            Priority: Minor
>
> We are storing messages on a per partition on disk. We can store each file (partition
data) on a different disk, if available, to take advantage of full IO bandwith, when using
multiple compute threads or workers per machine.

--
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