[rabbitmq-discuss] how to get higher throughput?

stone zmstone at gmail.com
Thu Mar 29 10:16:47 BST 2012


Thanks Alexis!
Just for the record,
I think I've hit the ceiling of the throughput ONE queue can provide,
because there is not difference between 12 and 24 clients in my tests.
Faster clients? not really the issue either, my workers just consume and
reply, do no extra computation.

Anyway, non-HA queue it's almost 5 times faster than HA in my tests.
I wasn't expecting such a significant difference.

BRs
/stone

On Thu, Mar 29, 2012 at 10:47 AM, Alexis Richardson <alexis at rabbitmq.com>wrote:

> HA replicates data to queue mirrors, which increases the amount of
> networking and data processing per message.  This can impact
> throughput.
>
> You may find you can get higher throughput by using more clients, or a
> faster client.
>
> alexis
>
>
> On Thu, Mar 29, 2012 at 9:35 AM, stone <zmstone at gmail.com> wrote:
> > Hi again.
> >
> > The traffic rate hit almost 9000/s if the "RPC" queue is non-HA.
> > Could someone explain a little bit why HA impact the throughput that
> much?
> >
> > BRs
> > /stone
> >
> >
> > On Wed, Mar 28, 2012 at 9:23 PM, stone <zmstone at gmail.com> wrote:
> >>
> >> Hi
> >>
> >> I have a use case which is very similar to RPC:
> >> The clients subscribe to their own queues bound to routing key
> >> "response.myname",
> >> and they publish messages with routing key "request.myname", then wait
> for
> >> response.
> >>
> >> On the other side of the broker, there is this RPC queue, bound to
> routing
> >> key "request.#",
> >> many workers consume from this queue, process the messages then send
> back
> >> response.
> >>
> >> The request size is around 10K bytes, response is just few bytes, and
> all
> >> "RPC"s are synchronized calls.
> >> i.e. client wait for response before send another request.
> >>
> >> After some benchmark tests, we found that the message rate of the RPC
> >> queue hits around 1400.
> >> It doesn't help by increasing the number of publishers or consumers, or
> >> both.
> >>
> >> Hope it's enough background info to finally ask my question:
> >> what could possibly be the bottleneck in my tests, and how to get higher
> >> throughput ?
> >>
> >> -----------------env-------------------
> >> Rabbit version: 2.7.1
> >> Memory high watermark: 25G
> >> CPU (24 processors) model: Intel(R) Xeon(R) CPU X5690  @ 3.47GH
> >>
> >> BRs
> >> /stone
> >>
> >
> >
> > _______________________________________________
> > 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/20120329/87ea079c/attachment.htm>


More information about the rabbitmq-discuss mailing list