[
https://issues.apache.org/jira/browse/HADOOP-10278?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13898660#comment-13898660
]
Chris Li commented on HADOOP-10278:
-----------------------------------
Interesting results from testing with different timeouts. Confirms previous
results, showing increase in client time alongside server time. Since I didn't
change any clientside code, it would seem that the client cpu time is dependent
on the server at least a bit.
One surprising result is that the 2500ms poll time fared a little bit worse
than 1000ms in throughput.
Test groups:
polltimeout_evenlonger with 39 samples - 2500ms poll time
polltimeout_longer with 38 samples - 1000ms poll time
trunk with 39 samples
{noformat}
=================
Test: client
=================
polltimeout_evenlonger vs polltimeout_longer
Changed by 0.23%, p=0.68354
*** polltimeout_evenlonger vs trunk
Changed by -1.05%, p=0.01338
*** polltimeout_longer vs trunk
Changed by -1.28%, p=0.02356
=================
Test: throughput
=================
*** polltimeout_evenlonger vs polltimeout_longer
Changed by -0.61%, p=0.02441
polltimeout_evenlonger vs trunk
Changed by -0.53%, p=0.09965
polltimeout_longer vs trunk
Changed by 0.07%, p=0.83207
=================
Test: server
=================
polltimeout_evenlonger vs polltimeout_longer
Changed by 0.30%, p=0.57614
*** polltimeout_evenlonger vs trunk
Changed by -1.99%, p=0.00016
*** polltimeout_longer vs trunk
Changed by -2.28%, p=0.00006
ttest assumes 2 sample same variance, 2 tailed.
*** marks where p value reaches significance (<0.05)
{noformat}
> Refactor to make CallQueue pluggable
> ------------------------------------
>
> Key: HADOOP-10278
> URL: https://issues.apache.org/jira/browse/HADOOP-10278
> Project: Hadoop Common
> Issue Type: Sub-task
> Components: ipc
> Reporter: Chris Li
> Attachments: HADOOP-10278-atomicref.patch,
> HADOOP-10278-atomicref.patch, HADOOP-10278-atomicref.patch, HADOOP-10278.patch
>
>
> * Refactor CallQueue into an interface, base, and default implementation that
> matches today's behavior
> * Make the call queue impl configurable, keyed on port so that we minimize
> coupling
--
This message was sent by Atlassian JIRA
(v6.1.5#6160)
