qpid-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Alex Rudyy (JIRA)" <j...@apache.org>
Subject [jira] [Commented] (QPID-8238) [Broker-J] Improve performance of asynchronous publishing of transient messages into topic exchange having queues bound using non-overlapping selectors
Date Tue, 11 Sep 2018 09:07:00 GMT

    [ https://issues.apache.org/jira/browse/QPID-8238?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16610324#comment-16610324

Alex Rudyy commented on QPID-8238:


Thanks a lot for the review comments. 

My changes for logback filter and fanout exchange routing functionality were driven by jprofiler
analysis indicating about performance being lost on unnecessary {{Map}} object creation. I
committed them against this JIRA as they improve the performance a bit whilst routing the
messages with fanout exchange but the performance of other exchange implementation should
improve as well. Though, I am happy to revert the mentioned changes and commit them against
separate JIRAs.

As for exchange changes, they not only optimize the code but make it much simpler. Effectively,
the changed code converts the collection of {{TopicMatcherResult}} got based on message routing
key into a map where keys are {{route destinations}} and values are {{routing/replacement
keys}}. Before my change the conversions was done by using intermediate map containing the
destination as a key and routing replacement key as a value. Than, multiple maps (representing
each {{TopicMatcherResult}} ) were merged into a single map.
My changes eliminated the need for intermediate map creation. I also do hope that code is
a bit simpler now. I left the old implementation untouched and marked is as {{Deprecated}}
in order to keep public methods unchanged as the next version is minor release rather major
one. (Though, I did not check whether we made other breakable changes in the code in general.
Perhaps, keeping dead-code there is a waist). I am planning to merge committed changes into
7.0 and do not want break backward compatibility there.

I liked  Rob's idea of making {{FieldTable}} immutable and do on-demand parsing. I will try
to implement the corresponding changes as part of this JIRA, however, making {{FieldTable}}
immutable most likely will not have any effect on performance and performance is the main
concern for me right now. I do not think I can port on-demand parsing changes into 7.0. Thus,
I am currently looking into other performance optimization alternatives. I have not looked
into on-demand parsing yet.

I played a bit with replacing of {{AMQShortString}}s with java {{String}} in {{BasicContentHeaderProperties}}
(I have a patch a corresponding changes). My performance tests did not revealed much performance
improvements due to using java {{String}} in {{BasicContentHeaderProperties}}. However, they
did improve the performance a bit. Potentially, making {{AMQShortString}}s to be a wrapper
of {{String}} and replacing usage of  {{AMQShortString}}s with java {{String}} where possible
could be another refactoring worth doing.

The current routing code is very well tuned. I do not see any other routing functionality
to optimize. As a next steps I am going to add new topic performance tests into existing perf
test suite and look into making {{FieldTable}} immutable and implementing on-demand parsing.

> [Broker-J] Improve performance of asynchronous publishing of transient messages into
topic exchange having queues bound using non-overlapping selectors 
> --------------------------------------------------------------------------------------------------------------------------------------------------------
>                 Key: QPID-8238
>                 URL: https://issues.apache.org/jira/browse/QPID-8238
>             Project: Qpid
>          Issue Type: Improvement
>          Components: Broker-J
>    Affects Versions: qpid-java-broker-7.0.6
>            Reporter: Alex Rudyy
>            Priority: Major
>             Fix For: qpid-java-broker-7.1.0, qpid-java-broker-7.0.7
>         Attachments: 0001-QPID-8238-Use-java.lang.String-for-keys-and-values-i.patch
> The performance of asynchronous publishing of transient messages into topic exchange
which routes messages into queues bound using non-overlapping selectors is 2-3 times slower
than performance of 0.32 broker. The performance degradation is observed with AMQP 0.9, though,
I suspect that the AMQP 0-10 protocol could be affected as well.
> I was running tests with 10 concurrent producers publishing messages  on separate connections
using the same routing key into 10 different queues (subscribers queues) bound to the exchange
using non-overlapping selectors.
> My testing showed that performance of 7.0 broker for this particular use case was 2-3
times worse than performance of 0.32 broker.
> The following factors contributed to degradation of performance:
> •	Copying data from direct memory into heap memory whilst decoding message headers.
Due to this factor, the decoding of message headers is around twice slower. It seems it contributes
around 70% to total performance degradation
> •	The message routing algorithm is slower due to need to support a new feature to route
messages into bound exchanges (in addition to queues) using replacement routing key.
> •	AMQ short strings caching contributes 5-10% to total performance degradation. The
caching was added to manage heap space more efficiently.
> The numbers provided here could be inaccurate due instrumentation overhead whilst profiling
the issue.
> Potentially, caching can be turned off but that will not improve performance much.
> On other hand, adding of additional caching of strings to amqp-short-strings would improve
the performance a bit. Whilst evaluating selectors, the fields used in selector expressions
are represented as java strings but they get converted every time into amqp-short-strings
when looking up for message header values. If 10 queues are bound to the exchange using the
same binding key, the selector expression is evaluated 10 times for the incoming message.
Thus, all selector field names are get converted into amqp-short-strings 10 times as well.
It seems adding caching here can improve the performance. 

This message was sent by Atlassian JIRA

To unsubscribe, e-mail: dev-unsubscribe@qpid.apache.org
For additional commands, e-mail: dev-help@qpid.apache.org

View raw message