[
https://issues.apache.org/jira/browse/HBASE-15492?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15205810#comment-15205810
]
Vladimir Rodionov commented on HBASE-15492:
-------------------------------------------
Applied 4 patches on top of HBASE-15479: HBASE-15493, HBASE-15491, HBASE-15501,
HBASE-15512. Set *hbase.hregion.memstore.chunkpool.maxsize=0.4* (HBASE-15513).
See attached 5patches.jfr.
Test run time 340 sec (down from 354 sec pre-HBASE-15479)
Total # of GC in young: 189 ( down from 360)
Total # of GC in old: 4 (down from 8)
Total time spent in GC: 6.7sec (down from 15 sec)
Averages and max GC times are also much better.
Total Allocation rate (combined TLAB + Old space) dropped from 1084MB/s to
704MB/sec (in reality it is even less because it reflects increased
throughput).
> Memory usage optimizations (Write path)
> ---------------------------------------
>
> Key: HBASE-15492
> URL: https://issues.apache.org/jira/browse/HBASE-15492
> Project: HBase
> Issue Type: Improvement
> Affects Versions: 2.0.0
> Reporter: Vladimir Rodionov
> Assignee: Vladimir Rodionov
> Fix For: 2.0.0
>
> Attachments: 5Patches.jfr, baseline.jfr, pre-HBASE-15479.jfr
>
>
> This is master ticket for all memory optimization tasks.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
