[rabbitmq-discuss] Task queues: use rabbitmq to build one, or use rabbitmq AS one
jerryk at rbcon.com
Tue Aug 14 00:19:47 BST 2012
As general background to the conversation, the following summarizes some
nice anti-patterns in the area:
These are good things to watch out for, since they almost spontaneously
self organize in a lot of systems we've seen. :-)
On Mon, Aug 13, 2012 at 2:38 PM, Mark Feeney <mark.feeney at gmail.com> wrote:
> Hi, Emile.
> Thanks for the response!
> > If you find yourself performing SELECT queries on the messages, and
> > their order is less important then consider using a relational database
> > instead.
> I do have some fairly tricky SELECT statements to get the next item
> form my current queue system, so this is a good point.
> > If not all messages are routed in an identical way, or you find yourself
> > writing lots of database trigger code then I would suggest using a
> > message broker. RabbitMQ can cope with long persistent queues, with
> > messages of variable sizes.
> This sounds like another good guideline. Right now I don't really
> have any message routing on the ingress side of my ad-hoc queue
> system. The SELECT statement that gets the next message ends up doing
> some fiddly logic to get the "correct" message out, and this can cause
> issues, esp. with larger queues. I'm considering routing messages to
> the correct queue when they first arrive, so rabbit seems like a good
> fit for this sort of thing.
> > If performance is a consideration then you should run some benchmarks
> > for your expected workload.
> Of course. Thanks again for your response, I'm getting a clearer
> picture of how I might use rabbitmq.
> rabbitmq-discuss mailing list
> rabbitmq-discuss at lists.rabbitmq.com
-------------- next part --------------
An HTML attachment was scrubbed...
More information about the rabbitmq-discuss