[rabbitmq-discuss] cpu & network usage

shridharan muthu shridharan.m at gmail.com
Wed Dec 4 23:11:16 GMT 2013

Hello there,

     As a performance optimization on our rabbit mq cluster (of 2 nodes), I
removed following from a high volume non-durable queue

   1. AMQP_MANDATORY flag during publish and
   2. AMQP_AUTOACK during consume.

 As a result the number of messages on that queue went down from 150 to 15
which is fantastic and expected. But this also increased overall cpu usage
from 22% to 32% with an increased network traffic from 30k packets/sec to
40k packets/sec. The ingress & egress rate stayed the same before and after
the change i.e. around 300 msgs/sec.

  So, my question is why would network traffic increase when I make all
calls asynchronous when the ingress & egress rate is same ?

More info about the cluster:
shri at mq1:~$ sudo rabbitmqctl status
Status of node rabbit at mq1 ...
     [{rabbitmq_management,"RabbitMQ Management Console","3.0.2"},
      {rabbitmq_management_agent,"RabbitMQ Management Agent","3.0.2"},
      {os_mon,"CPO  CXC 138 46","2.2.4"},
      {rabbitmq_mochiweb,"RabbitMQ Mochiweb Embedding","3.0.2"},
      {mochiweb,"MochiMedia Web Server","2.3.1-rmq3.0.2-gitd541e9a"},
      {mnesia,"MNESIA  CXC 138 12","4.4.12"},
      {amqp_client,"RabbitMQ AMQP Client","3.0.2"},
      {xmerl,"XML parser","1.2.3"},
      {inets,"INETS  CXC 138 49","5.2"},
      {sasl,"SASL  CXC 138 11","2.1.8"},
      {stdlib,"ERTS  CXC 138 10","1.16.4"},
      {kernel,"ERTS  CXC 138 10","2.13.4"}]},
     "Erlang R13B03 (erts-5.7.4) [source] [64-bit] [smp:12:12] [rq:12]
[async-threads:30] [hipe] [kernel-poll:true]\n"},

Thank you,
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.rabbitmq.com/pipermail/rabbitmq-discuss/attachments/20131204/15110bbd/attachment.html>

More information about the rabbitmq-discuss mailing list