[
https://issues.apache.org/jira/browse/HADOOP-3596?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=12611916#action_12611916
]
Qi Ji commented on HADOOP-3596:
-------------------------------
Yeah, when the file number approximates 100 million the overall performance is
greatly affected by name node performance. We are going to work this out by a
patch about cooperative multi namenodes soon.
> Namenode cluster(multi namenode working in active-active mode) to keep high
> availability and expansivity
> --------------------------------------------------------------------------------------------------------
>
> Key: HADOOP-3596
> URL: https://issues.apache.org/jira/browse/HADOOP-3596
> Project: Hadoop Core
> Issue Type: New Feature
> Components: dfs
> Environment: Hadoop cluster on more than thousands of nodes, most of
> all, the number of files is mass.
> Reporter: Meng Xu
> Original Estimate: 672h
> Remaining Estimate: 672h
>
> If we use the DFS to store a great quantity of files which is very small,
> i.e. the namenode is under high pressure of indexing files, should we
> consider to introduce other namenodes working concurrently for availability
> and scalability?
--
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.
