[rabbitmq-discuss] Question on throughput with RabbitMQ-3.1.1

Priyanki Vashi vashi.priyanki at gmail.com
Wed Jun 26 18:52:19 BST 2013


Hi again,

When I changed the connection from blocking to non-blocking (using select
connection method) my throughput has increased hugely.
Also correctly configured publisher confirm as per yesterday's suggestions.

I am yet to try different things but from the initial tests on high end
server now I can see I can publish 8000 msg/Sec with single publisher and
no consumer.
On my macbook I get 1800 msg/sec throughput with single producer and single
consumer.

Thank you all again. It was very helpful for my work. I will not try to
simulate multiple producers/consumers using this improvements and see what
throughout I get.

//P


On Tue, Jun 25, 2013 at 6:39 PM, PRIYANKI VASHI <vashi.priyanki at gmail.com>wrote:

> Hi ,
>
> Thanks Micheal and everyone else.
> I got some good insight now on my problems.
>
> Will try to do some changes based on them and see how it goes.
>
> Will write back.
>
> //P
>
> Sent from my iPhone
>
> On 25 jun 2013, at 18:20, Michael Klishin <michael.s.klishin at gmail.com>
> wrote:
>
>
> 2013/6/25 Priyanki Vashi <vashi.priyanki at gmail.com>
>
>> What is the concept of blocking connection ?
>> What difference really it will make over other type of connection ?
>>
>> Some theory on this would be helpful.
>>
>
> Blocking I/O-based connection assumes the caller blocks (waits) until the
> response arrives.
>
> With a non-blocking implementation, the caller does not wait, which means
> your program can write
> continuously but also that responses that arrive at a later point are
> handled differently, often
> forcing developers to use less familiar APIs and manage program state
> differently.
>
> Pika offers you to choose from multiple connection implementations. Michael
> suggests you to start with the most straightforward one and when you have
> that working,
> switch to a non-blocking implementation (e.g. Tornado-based) which often
> yields higher
> throughput.
>
> http://en.wikipedia.org/wiki/Asynchronous_I/O
> http://www.sal.ksu.edu/faculty/tim/ossg/Device/blocking.html
> --
> MK
>
> http://github.com/michaelklishin
> http://twitter.com/michaelklishin
>
> _______________________________________________
>
> rabbitmq-discuss mailing list
> rabbitmq-discuss at lists.rabbitmq.com
> https://lists.rabbitmq.com/cgi-bin/mailman/listinfo/rabbitmq-discuss
>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.rabbitmq.com/pipermail/rabbitmq-discuss/attachments/20130626/158629eb/attachment.htm>


More information about the rabbitmq-discuss mailing list