hbase-issues mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "ramkrishna.s.vasudevan (JIRA)" <j...@apache.org>
Subject [jira] [Updated] (HBASE-12213) HFileBlock backed by Array of ByteBuffers
Date Thu, 02 Jul 2015 11:59:05 GMT

     [ https://issues.apache.org/jira/browse/HBASE-12213?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]

ramkrishna.s.vasudevan updated HBASE-12213:
-------------------------------------------
    Status: Patch Available  (was: Open)

Uses MBB in the read path.  
-> Since HBASE-12295 is not there we cannot allow the Bucket Cache to use the actual MBB
based Blocks without doing copy. For now we have copied the Bucketcache and created an MBB
out of it.
-> the HFileReader uses MBB and the relative reads have been replaced with absolute position
based reads.  
The absolute position based reads uses the UnsafeAccess APIs. So better to make use of them.
Tried to do a small micro benchmark mimicing the logic in blockSeek with and without positional
reads using MBB.
-> There are some TODOs that will get changed after BufferBackedCells come in to place.
 
-> PrefixTree, blooms need to be handled to work with MBB that are offheap.  That can be
done later.

> HFileBlock backed by Array of ByteBuffers
> -----------------------------------------
>
>                 Key: HBASE-12213
>                 URL: https://issues.apache.org/jira/browse/HBASE-12213
>             Project: HBase
>          Issue Type: Sub-task
>          Components: regionserver, Scanners
>            Reporter: Anoop Sam John
>            Assignee: ramkrishna.s.vasudevan
>         Attachments: HBASE-12213_1.patch
>
>
> In L2 cache (offheap) an HFile block might have been cached into multiple chunks of buffers.
If HFileBlock need single BB, we will end up in recreation of bigger BB and copying. Instead
we can make HFileBlock to serve data from an array of BBs.



--
This message was sent by Atlassian JIRA
(v6.3.4#6332)

Mime
View raw message