[
https://issues.apache.org/jira/browse/HBASE-15756?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
stack updated HBASE-15756:
--------------------------
Attachment: patched.vs.patched_and_cached.vs.no_patch.png
Thanks for the patch [~aoxiang] (I didn't realize it had to go as last option
on bootstrap -- thanks).
Here is a graph with three peaks. The first is with the patch in place. Looking
in jstacks, allocations no longer show so we must be getting from the pool now.
We seem to be close to the unpatched version but slightly slower still.
Turns out all data has not been fitting in cache (I'd messed w/ configs). I
made the configs so all data does fit and when I run this way, I see that the
netty server is a good bit slower than what we have: 300k ops/sec vs 400k
ops/sec... This is after my constraining netty not to run away with workers
making it so only 6.
I'll try and look and see if anything obvious in the difference but maybe you
have ideas [~aoxiang]? Thanks.
> Pluggable RpcServer
> -------------------
>
> Key: HBASE-15756
> URL: https://issues.apache.org/jira/browse/HBASE-15756
> Project: HBase
> Issue Type: Improvement
> Components: Performance, rpc
> Reporter: binlijin
> Assignee: binlijin
> Priority: Critical
> Attachments: Netty4RpcServer_forperf.patch, NettyRpcServer.patch,
> NettyRpcServer_forperf.patch, PooledByteBufAllocator.patch,
> PooledByteBufAllocator2.patch, gc.png, gets.png, gets.png, idle.png,
> patched.vs.patched_and_cached.vs.no_patch.png, queue.png
>
>
> Current we use a simple RpcServer, and can not configure and use other
> implementation.This issue is to make the RpcServer pluggable, so we can make
> other implementation for example netty rpc server. Patch will upload laterly
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
