[
https://issues.apache.org/jira/browse/HBASE-20188?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16404441#comment-16404441
]
Anoop Sam John commented on HBASE-20188:
----------------------------------------
I see.. no worries Yu.. I was just speaking from the latest comments point.
Ya in the tests the change is aysn wal and the compacting memstore. Off heap
is not ON by default and I dont think stack did that in his tests.
Ya 2 aspects. The slower write perf compared to 1.x And why getting so many
Region busy Exceptions.
Enabling off heaping would help us for having a smaller heap size and so GC
advantages. But in tests we did not see that big gain as what we saw in read
path off heaping. The read side was like 20%+ avg throughput gain and here it
was like <10% on write throughput. I dont remember the exact numbers. The off
heaping of write path was our stepping stones for HBASE-20003
> [TESTING] Performance
> ---------------------
>
> Key: HBASE-20188
> URL: https://issues.apache.org/jira/browse/HBASE-20188
> Project: HBase
> Issue Type: Umbrella
> Components: Performance
> Reporter: stack
> Priority: Critical
> Fix For: 2.0.0
>
> Attachments: flamegraph-1072.1.svg, flamegraph-1072.2.svg, tree.txt
>
>
> How does 2.0.0 compare to old versions? Is it faster, slower? There is rumor
> that it is much slower, that the problem is the asyncwal writing. Does
> in-memory compaction slow us down or speed us up? What happens when you
> enable offheaping?
> Keep notes here in this umbrella issue. Need to be able to say something
> about perf when 2.0.0 ships.
--
This message was sent by Atlassian JIRA
(v7.6.3#76005)
