A little more CPU usage compared to no UUID namespace configuration at all, or
compared to the old temp. queue namespace? To correct my previous statement, we
*migrated* from temp. namespace to new UUID namespace configuration.
--
Vilius
-----Original Message-----
From: Justin Bertram <[email protected]>
Sent: Friday, October 31, 2025 3:11 PM
To: [email protected]
Subject: Re: Performance impact of connection router
There will be a little more CPU usage when using the uuid-namespace since the
broker has to detect the UUID, but again, I wouldn't expect that to be
significant.
Justin
On Fri, Oct 31, 2025 at 1:58 AM Vilius Šumskas
<[email protected]> wrote:
> Oh, and one more thing, we also started using the new UUID namespace
> configuration and security rules in 2.42.0. But I guess that should
> not be a problem performance wise(?). Most of our connections use
> temporary UUID queues to reply to.
>
> --
> Vilius
>
> -----Original Message-----
> From: Vilius Šumskas <[email protected]>
> Sent: Friday, October 31, 2025 8:51 AM
> To: [email protected]
> Subject: RE: Performance impact of connection router
>
> I didn't at the moment, but I will try to change configuration on our
> production system during next maintenance window to see how it affects
> CPU utilization.
>
> The reason I'm asking is that I started to investigate why one of our
> clusters has a considerable CPU usage, even though the throughput load
> (message count and size) is not that big at all. It is either:
> a) a connection router configuration we recently introduced, or
> b) something changed on the application side, maybe how developers
> close and open connections, or something like that.
>
> --
> Vilius
>
> -----Original Message-----
> From: Justin Bertram <[email protected]>
> Sent: Friday, October 31, 2025 5:18 AM
> To: [email protected]
> Subject: Re: Performance impact of connection router
>
> The performance impact is hard to judge, but I wouldn't expect it to
> be much as it really just adds one invocation of
> java.util.regex.Matcher#matches for every connection.
>
> Have you conducted performance tests with and without the router? If
> so, what did you find?
>
>
> Justin
>
> On Thu, Oct 30, 2025 at 6:12 PM Vilius Šumskas
> <[email protected]>
> wrote:
>
> > Sorry last message went unfinished to the list.
> >
> > ---
> >
> > We are using Artemis router:
> > <connection-router name="deny-privileged-users">
> > <key-type>ROLE_NAME</key-type>
> > <local-target-filter>^(?!amq$).*$</local-target-filter>
> > </connection-router>
> >
> > to filter out unwanted users from connecting to one of our acceptors.
> >
> > I'm wondering, what is the performance impact of such router if we
> > have large amount of non-persistent connections from the clients? We
> > have ~3k connections at the moment with a fair amount of them
> > constantly disconnecting and connecting as per business requirements.
> > Connections are TLS protected, if that make a difference.
> >
> > --
> > Vilius
> >
> > -----Original Message-----
> > From: Vilius Šumskas <[email protected]>
> > Sent: Friday, October 31, 2025 1:03 AM
> > To: [email protected]
> > Subject: Performance impact of acceptor router
> >
> > Hi,
> >
> > we are using router:
> >
> > --
> > Pagarbiai,
> >
> > Vilius Šumskas
> > Rivile
> > IT vadovas
> > +370 614 75713
> >
> >
> > --------------------------------------------------------------------
> > - To unsubscribe, e-mail: [email protected]
> > For additional commands, e-mail: [email protected] For
> > further information, visit: https://activemq.apache.org/contact
> >
> >
>
> ---------------------------------------------------------------------
> To unsubscribe, e-mail: [email protected]
> For additional commands, e-mail: [email protected] For
> further information, visit: https://activemq.apache.org/contact
>
>
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
For further information, visit: https://activemq.apache.org/contact
