Since you are comparing to a previous integration tool ; was previous
integration tool using async option with another queue within
"integration"?

ચિરાગ/चिराग/Chirag
------------------------------------------
Sent from My Gmail Account


On Sat, Sep 27, 2025 at 11:13 AM Peter Hicks
<[email protected]> wrote:

> Salut Gael
>
> On Monday, 22 September 2025 at 18:13, Gael LE BELLEGO <
> [email protected]> wrote:
>
> > The routes are going well... but for one case : when we consume from or
> publish to MqSeries, we end up having terrible performances.
> > On a previous application, (implemented with another EIP framework), we
> had a mean message rate of 40 m/s with only one consumer or producer.
> > Currently, our message rate dropped to 2 m/s with the JMS component.
> > (similar comparing conditions).
>
> I have experience of using IBM MQ (not heard it called MQSeries for
> years!) and I can offer you these suggestions:
>
> 1. Switch to using the MQ 'allclient' directly and compare performance -
> in the past I've had problems where publish or subscribe rates were limited
> because a client was connecting, sending/receiving, then disconnecting.
> 'tcpdump' can help you identify if your client is reusing or setting up a
> new TCP connection as new connections are expensive in MQ terms
> 2. The overhead on persistent messages with IBM MQ is higher than I
> thought it would be, so check your performance when publishing -
> particularly, the I/O throughput of the server is important here
>
> I've been running Camel inside ActiveMQ for 8+ years consuming from an MQ
> Server and republishing to a local AMQ topic - it works really really well.
>
>
> Peter
>

Reply via email to