<span style="background-color: transparent; ">On Tue, Nov 8, 2011 at 11:11 AM, Matthias Radestock </span><span dir="ltr" style="background-color: transparent; "><<a href="mailto:matthias@rabbitmq.com">matthias@rabbitmq.com</a>></span><span style="background-color: transparent; "> wrote:</span><br>
<div class="gmail_quote"><blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex;"><div class="im">On 08/11/11 18:22, Al Tobey wrote:<br>
<blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex">
1.) We started paying attention and caring about delays<br>
</blockquote>
<br></div>
If messages were "going in and coming out just fine at 5000-6000 messages/s", then no backlog, and hence no increasing delay, can have developed.<br>
<br>
If rabbit was happily coping with 5-6kHz at one point, without creating a backlog, but later on it was only managing 1.2kHz then one possible explanation is that a temporary spike may have caused enough of a backlog to build up to force rabbit to page to disk, at which point performance drops until the backlog clears.</blockquote>
<div><span style="background-color: transparent; "><br></span></div><div><span style="background-color: transparent; ">Sorry, I wasn't clear or thorough. I don't know if they were backing up or not, because I turned it all up and moved on to other things. It might have been backing up or delayed without tipping the server over; I wasn't watching until the engineers using it notified me that they were seeing delay.</span></div>
<div><span style="background-color: transparent; "><br></span></div><blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex;"><div class="im">
<blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex">
2.) Load increased, by as much as 2x depending on time of day<br>
</blockquote>
<br></div>
So is your main complaint then that the message rates rabbit can cope with when using stomp are lower than what you expect to see? If so, what is that expectation based on? A comparison with using RabbitMQ with AMQP? Rabbit's stomp processing is a fair bit slower than handling AMQP.<br>
</blockquote><div><br></div><div>Yeah, that's my complaint (really, more of an observation). I would hope it could be within a few percent of AMQP or even faster since the protocol is smaller & simpler, but I totally understand why RabbitMQ chose to back it on AMQP.</div>
<div><span style="background-color: transparent; "><br></span></div><div><span style="background-color: transparent; ">I'm pretty sure I've pushed way more messages through RabbitMQ+STOMP before, but that was a couple years ago at another company, and also on rather nice bare metal so it's an unfair comparison.</span></div>
<div><br></div><div>"Rabbit's stomp processing is a fair bit slower than handling AMQP" is what I suspected and is a satisfactory answer.</div><div><br></div><div>Thanks,</div><div>-Al</div><div> </div><blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex;">
Regards,<br><font color="#888888">
<br>
Matthias.<br>
</font></blockquote></div><br>