[rabbitmq-discuss] [rabbitmq-users] Re: Can RabbitMQ handle big messages?

Wed Aug 6 16:57:10 BST 2014

On 6 August 2014 at 19:52:53, Laing, Michael (michael.laing at nytimes.com) wrote:
> > Based upon our own benchmarks, and common sense, we store large 
> message bodies externally (we use s3 but there are lots of solutions) 
> and pass a reference in a header.

Yup, this is a strategy that is known to work well.

> It's pretty easy to generate a signed URL with a one-off key that 
> can be particularized to a specific user or audience and given 
> a ttl from a few seconds to a few days.

Several K/V or blob stores (Cassandra, Riak, Riak CS, Atmos, LeoFS) could
be quite effective at this, too.
-- 
MK 

Staff Software Engineer, Pivotal/RabbitMQ