[
https://issues.apache.org/jira/browse/HDFS-5453?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13853091#comment-13853091
]
Milind Bhandarkar commented on HDFS-5453:
-----------------------------------------
Mem-to-Mem replication is easier to do, and is already norm in several
distributed data systems. Plus RoCE gives us opportunity to take CPU out of the
equation, by giving the network more intelligence in doing such replication,
across a wide network. Given the steep downtrending cost-curves, I believe it
will be widespread deployment. My question is, as I asked more than 2 years
ago, the same. Are you amenable to rapidly coming future technology and
economics, or going to be stuck in the past by giving reasons about rotating
disk arm latency?
> Support fine grain locking in FSNamesystem
> ------------------------------------------
>
> Key: HDFS-5453
> URL: https://issues.apache.org/jira/browse/HDFS-5453
> Project: Hadoop HDFS
> Issue Type: New Feature
> Components: namenode
> Affects Versions: 2.0.0-alpha, 3.0.0
> Reporter: Daryn Sharp
> Assignee: Daryn Sharp
>
> The namesystem currently uses a course grain lock to control access. This
> prevents concurrent writers in different branches of the tree, and prevents
> readers from accessing branches that writers aren't using.
> Features that introduce latency to namesystem operations, such as cold
> storage of inodes, will need fine grain locking to avoid degrading the entire
> namesystem's throughput.
--
This message was sent by Atlassian JIRA
(v6.1.4#6159)
