185958 (4) [Avatar] Offline
#1
Hello, Great book, well done. I learn a lot!

Some ideas on ch.6...

On performance / section 6.5 and hashing 6.7, the analysis looks a bit weird. What I would conclude is:

On Figure 6.13: The larger the message you send, the slower the queue. It's not related to headers specifically. It's just that the payload/message is so small that your headers end up being a large percentage of the overall message size.

Figure 6.14: "t’s clear that the topic exchange is more performant than the headers exchange" <- I don't think there's anything clear there. The only clear thing is that performance is not an issue. Just focus on the use case and use the simplest thing possible.

On Figure 6.17 ... it doesn't look like a small discrepancy. That's a good 15% less on q0. I think you should investigate this a bit further - it's really weird! I tried to reproduce it but I used rabbitmq 3.6.6. I had to do
sudo rabbitmq-plugins enable rabbitmq_consistent_hash_exchange
but
rabbitpy
didn't want to connect with it with traces like this:


Traceback (most recent call last):
  File "6.7 Creating Multiple Bound Queues.py", line 10, in <module>
    queue.bind('image-storage', '10')
  File "/home/ubuntu/.local/lib/python2.7/site-packages/rabbitpy/connection.py", line 149, in __exit__
    raise exc_val
rabbitpy.exceptions.AMQPNotFound: <pamqp.specification.Channel.Close object at 0x7f890eb92fa0>


I tried to use custom URL etc but didn't finish it.