[
https://issues.apache.org/jira/browse/ARTEMIS-3045?focusedWorklogId=528110&page=com.atlassian.jira.plugin.system.issuetabpanels:worklog-tabpanel#worklog-528110
]
ASF GitHub Bot logged work on ARTEMIS-3045:
-------------------------------------------
Author: ASF GitHub Bot
Created on: 24/Dec/20 12:20
Start Date: 24/Dec/20 12:20
Worklog Time Spent: 10m
Work Description: franz1981 edited a comment on pull request #3392:
URL: https://github.com/apache/activemq-artemis/pull/3392#issuecomment-750867139
@michaelandrepearce that's a PR that aim to "fix" the regression seen on
https://softwaremill.com/mqperf/
TBH I just reproduced the scalability issue with small message size given
that we tends to underutilized the network because we sent replicated packets
ASAP without batching them (that means that a single standard MTU of 1500 bytes
could contains x10 tiny packets eg ack, small records..).
This PR batch replicated packets by filling the Netty pending buffer and
it's getting a nice speed up if compared with master (and 2.2.0, surprisingly):
master:
```
**************
RUN 3 EndToEnd Throughput: 20447 ops/sec
**************
EndToEnd SERVICE-TIME Latencies distribution in MICROSECONDS
mean 1635.00
min 180.22
50.00% 1523.71
90.00% 2088.96
99.00% 4521.98
99.90% 8028.16
99.99% 13959.17
max 17825.79
count 320000
```
This PR:
```
**************
RUN 4 EndToEnd Throughput: 24524 ops/sec
**************
EndToEnd SERVICE-TIME Latencies distribution in MICROSECONDS
mean 1310.83
min 182.27
50.00% 1204.22
90.00% 1736.70
99.00% 3850.24
99.90% 7995.39
99.99% 12713.98
max 19267.58
count 320000
```
this is still a DRAF PR given that we need further tests to see that's not
regressing perf for other cases and I'm open to experiment other options.
The work on this PR is directly related with
https://issues.apache.org/jira/browse/ARTEMIS-2877
----------------------------------------------------------------
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
For queries about this service, please contact Infrastructure at:
[email protected]
Issue Time Tracking
-------------------
Worklog Id: (was: 528110)
Time Spent: 0.5h (was: 20m)
> ReplicationManager can batch sent replicated packets
> ----------------------------------------------------
>
> Key: ARTEMIS-3045
> URL: https://issues.apache.org/jira/browse/ARTEMIS-3045
> Project: ActiveMQ Artemis
> Issue Type: Improvement
> Reporter: Francesco Nigro
> Assignee: Francesco Nigro
> Priority: Major
> Time Spent: 0.5h
> Remaining Estimate: 0h
>
--
This message was sent by Atlassian Jira
(v8.3.4#803005)
