[
https://issues.apache.org/jira/browse/CASSANDRA-2252?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13082909#comment-13082909
]
Jonathan Ellis commented on CASSANDRA-2252:
-------------------------------------------
bq. Adding region recycling post-flush
Of course this has the same reference counting problem as JNA free does: we
can't re-use the buffer until we're sure nobody's reading from it anymore.
I'm inclined to just do in-heap buffers for the memtables to start with.
Unlike cached rows, we're virtually guaranteed contention on the memtable
objects if we try to do AtomicInteger-based reference counting, and sharded
threadlocal counting is more of a pita than I want to tackle at the moment.
I suspect that, like HBase, we'll realize substantial wins from area allocation
even on-heap.
> off-heap memtables
> ------------------
>
> Key: CASSANDRA-2252
> URL: https://issues.apache.org/jira/browse/CASSANDRA-2252
> Project: Cassandra
> Issue Type: Improvement
> Reporter: Jonathan Ellis
> Assignee: Jonathan Ellis
> Fix For: 1.0
>
> Attachments: 0001-add-MemtableAllocator.txt,
> 0002-add-off-heap-MemtableAllocator-support.txt, 2252-v3.txt, merged-2252.tgz
>
> Original Estimate: 0.4h
> Remaining Estimate: 0.4h
>
> The memtable design practically actively fights Java's GC design. Todd
> Lipcon gave a good explanation over on HBASE-3455.
--
This message is automatically generated by JIRA.
For more information on JIRA, see: http://www.atlassian.com/software/jira
