[
https://issues.apache.org/jira/browse/HBASE-11710?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14175830#comment-14175830
]
stack commented on HBASE-11710:
-------------------------------
We need a pread version of the HasEnhancedByteBufferAccess zero-copy methods.
As is, need to seek then BB read but the seek and BB read must be under a
lock/synchronize. Won't work for us unless we do the pread version (each works
on its own block... no synchronizing on stream..)
> if offheap L2 blockcache, read from DFS into offheap and copy to blockcache
> without coming onheap
> -------------------------------------------------------------------------------------------------
>
> Key: HBASE-11710
> URL: https://issues.apache.org/jira/browse/HBASE-11710
> Project: HBase
> Issue Type: Brainstorming
> Components: BlockCache, Offheaping
> Reporter: stack
> Assignee: stack
>
> Filing issue assigned to myself to take a look at feasibility of our doing
> this
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)