[
https://issues.apache.org/jira/browse/HBASE-10719?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13931471#comment-13931471
]
Matt Corgan commented on HBASE-10719:
-------------------------------------
Right, all current data block encoders leave values fully intact as a single
byte range. Values only need to be identified by a reference with
offset/length and that reference can be passed up the chain until it's copied
over the network. They may currently get copied into a KeyValue byte[], but
that won't be necessary if we (i guess you, Ram =) can get Cells passed all the
way up the read path. My hunch would be to get Cells all the way up the read
path and then go back and replace Cell.getValueArray() with
Cell.getValueBuffer(), but you know the code far better than me.
> Cell interface may need a ByteBuffer rather than a byte[]
> ---------------------------------------------------------
>
> Key: HBASE-10719
> URL: https://issues.apache.org/jira/browse/HBASE-10719
> Project: HBase
> Issue Type: Sub-task
> Reporter: ramkrishna.s.vasudevan
> Assignee: ramkrishna.s.vasudevan
>
> See discussion in HBASE-10531.
> The byte[] backing a cell may be off heap also. In such a case having a
> byte[] may require a copy offheap to on heap. So better to have ByteBuffer
> backing a Cell.
--
This message was sent by Atlassian JIRA
(v6.2#6252)
