Github user franz1981 commented on the issue:
https://github.com/apache/activemq-artemis/pull/945
@jbertram
I've attached the first results on
[localhost.txt](https://github.com/apache/activemq-artemis/files/687370/localhost.txt).
The throughputs (end to end) are pretty the same (a lil' adavantage for the
JCtools maps) and the memory footprint is similar.
What is pretty different is the latency distribution: the JCtools version
deliver a more stable throughput probably due to fewer and shorter GC pauses
(there are less Long and Node instances produced and collected).
The test configuration is:
- Max throughput, single queue, concurrent send and receive
- Non-persistent messages, at-least-once delivery, auto acknowledge (JMS)
- same machine RHEL 6.9 OpenJDK 1.8.0
- 100000 warmup messages + 5 runs of 100000 messages
- 100 bytes per message
I've tested everything on the same machine with the default (AIO)
persistence and the Artemis JMS client.
It is not the typical benchmark that helps the JCtools maps to shine
against the vanilla one.
I'm building a micro-benchmark to stress the journal and provides best case
results too :)
---
If your project is set up for it, you can reply to this email and have your
reply appear on GitHub as well. If your project does not have this feature
enabled and wishes so, or if the feature is enabled but not working, please
contact infrastructure at [email protected] or file a JIRA ticket
with INFRA.
---
