[
https://issues.apache.org/jira/browse/HBASE-14613?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14959381#comment-14959381
]
Enis Soztutar commented on HBASE-14613:
---------------------------------------
Echoing my comments from the other thread:
Do we know why is this off by default? Chunks are always equal in size, so
re-using the chunks makes a lot of sense. Otherwise every new chunk allocation
has to go through the eden space until it is tenured causing un-necessary work
for the GC. Should we look into enabling the chunk pool by default instead?
Todd and Lars G pointed out that the allocation may already be happening from
the tenured space, so no extra work for the GC.
https://github.com/fpavageau/tenured-allocation says that it is >2MB for CMS
and 2MB for G1. I did not run the test myself.
> Remove MemStoreChunkPool?
> -------------------------
>
> Key: HBASE-14613
> URL: https://issues.apache.org/jira/browse/HBASE-14613
> Project: HBase
> Issue Type: Brainstorming
> Reporter: Lars Hofhansl
> Priority: Minor
> Attachments: 14613-0.98.txt
>
>
> I just stumbled across MemStoreChunkPool. The idea behind is to reuse chunks
> of allocations rather than letting the GC handle this.
> Now, it's off by default, and it seems to me to be of dubious value. I'd
> recommend just removing it.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
