[
https://issues.apache.org/jira/browse/HBASE-15756?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
stack updated HBASE-15756:
--------------------------
Attachment: queue.png
idle.png
gets.png
I tried the patch. There are three runs: current state of branch-1, then
patch-as-is, and then patch with max i/o workers of 6. The latter change was to
make it so netty config more resembled the first runs config (6 reader threads).
Looking at the charts, there is a bit of an improvement. We are doing 110k
ops/second vs about 105k ops/second. This is not a large difference but at the
same time, given how little of an incursion has been made -- just swapping the
Listern/Readers for netty boss thread and 'i/o workers' -- it is pretty nice.
No tuning was done.
How hard to do a netty4 [~aoxiang] in here? I could try that and we could just
switch?
What you think of the questions [~anoop.hbase] and [~ram_krish] are asking?
It'd be critical that the rpcserver does not get in the way of their offheaping
(offheap netty BBuff as far as the MSLAB blocks?)
> 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: NettyRpcServer.patch, NettyRpcServer_forperf.patch,
> gets.png, idle.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)
