[
https://issues.apache.org/jira/browse/HDFS-14401?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16809519#comment-16809519
]
Anoop Sam John commented on HDFS-14401:
---------------------------------------
Now the new config "dfs.datanode.cache.loader.class" is mandatory to use the
pmem based cache. I know this was added because we have the Java based impl and
another subtack is planning for a native impl (based on availability of PMDK
lib)
Being a user I think it is an unwanted overhead. When we have a native impl and
pure Java impl can HDFS select which loader to be used automatically? If the
native lib is available in a node, the native loader only can be used? The
native impl is added because it is much better performing for reads and writes
from/to cache. So when the node is ready for native loader and that is a
better performing one, I am not sure why being a user I should try to use the
less performing loader version.
> Refine the implementation for HDFS cache on SCM
> -----------------------------------------------
>
> Key: HDFS-14401
> URL: https://issues.apache.org/jira/browse/HDFS-14401
> Project: Hadoop HDFS
> Issue Type: Sub-task
> Components: caching, datanode
> Reporter: Feilong He
> Assignee: Feilong He
> Priority: Major
>
> In this Jira, we will refine the implementation for HDFS cache on SCM, such
> as: 1) Handle full pmem volume in VolumeManager; 2) Refine pmem volume
> selection impl; 3) Clean up MapppableBlockLoader interface; etc.
--
This message was sent by Atlassian JIRA
(v7.6.3#76005)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
