[rabbitmq-discuss] Producer consumer performance test

Shalin Shekhar Mangar shalinmangar at gmail.com
Thu Apr 21 21:21:41 BST 2011


On Thu, Apr 21, 2011 at 3:39 PM, Matthias Radestock
<matthias at rabbitmq.com>wrote:

> Shalin,
>
> Shalin Shekhar Mangar wrote:
>
>> The consumption rate holds throughout the test but it still seems low
>> to me. I'd think that for such a small message payload, a single box
>> should be able to do much more than these.
>>
>
> Try running the MulticastMain example that ships with the Java client,
> using the "-a -r <rate>" option and set the rate to 20000. That should be
> stable, i.e. production and consumption rates should match. Depending on
> your machine, you can probably go a bit higher than that.
>
>
I'll try that out.


>
>  Is there something else that I should do to get better performance
>> out of a single box?
>>
>
> For a single producer, queue, consumer scenario the amount of parallelism
> is limited, so rabbit won't be utilising all the cores. If you have multiple
> independent streams, and/or fanout the figures will be higher.
>
>
I see. Let me try this out. Thanks!

-- 
Regards,
Shalin Shekhar Mangar.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.rabbitmq.com/pipermail/rabbitmq-discuss/attachments/20110421/a3ef6892/attachment-0001.htm>


More information about the rabbitmq-discuss mailing list