[rabbitmq-discuss] RabbitMQ 3.0 Policy ha-all delete issue
ward.mark at gmail.com
Mon Jan 14 19:35:53 GMT 2013
I have been on other projects and getting back to thinking about RabbitMQ
and what you said
>That would be the "one queue per message" anti-pattern.
Would the one-queue-per-message still be an anti-pattern if the cost of
creating queues was less?
I have tried to find more information about this anti-pattern. I am
thinking it is an anti-pattern due to the cost to create, subscribe, and
publish the queue on a per message basis. I am not finding information if
it is a messaging architecture anti-pattern.
I went with the one-queue-per-message for the request/response model to
avoid single-purpose queues with a listener class that acted as a router
between the requester and the queue. No two thread request/response calls
should block each other due to being sent to the same queue.
You have provided good ideas for working around performance issues with
short-lived queues. Right now I am trying to understand the anti-pattern
View this message in context: http://rabbitmq.1065348.n5.nabble.com/RabbitMQ-3-0-Policy-ha-all-delete-issue-tp23527p24444.html
Sent from the RabbitMQ mailing list archive at Nabble.com.
More information about the rabbitmq-discuss