You are viewing a plain text version of this content. The canonical link for it is here.
Posted to dev@hama.apache.org by "Hyunsik Choi (JIRA)" <ji...@apache.org> on 2010/02/22 14:16:27 UTC

[jira] Commented: (HAMA-199) Add fault tolerance to BSPPeer

    [ https://issues.apache.org/jira/browse/HAMA-199?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=12836613#action_12836613 ] 

Hyunsik Choi commented on HAMA-199:
-----------------------------------

This issue is really important because our one aims at large scale processing that takes long time. If hama doesn't have this feature and at least one server is fault, we have to restart that processing.

Basically, this feature would be good to follow the fault tolerance model of the mapreduce, materializing immediate results into local storage. Actually, that model is not efficient, but that is robust. Later, we will improve this feature to be pipelined like mapreduce online.

> Add fault tolerance to BSPPeer
> ------------------------------
>
>                 Key: HAMA-199
>                 URL: https://issues.apache.org/jira/browse/HAMA-199
>             Project: Hama
>          Issue Type: Improvement
>          Components: bsp
>    Affects Versions: 0.2.0
>            Reporter: Hyunsik Choi
>            Assignee: Hyunsik Choi
>
> This version of BSPPeer has no fault tolerance ability. When one or more BSPPeers lead to fault, other BSPPeers will hang. So, BSPPeer needs to be fault tolerant.

-- 
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.