This is not a question of preventing patent trolls from patenting the same thing.
Firstly, because AMQP has hundreds or thousands of areas that could be similarly patented: failover, federation, many types of exchange, remote administration, etc. It only takes one patent to hold the whole standard to ransom. Red Hat would have to patent every single technical aspect of the standard, which would be impossible in practical terms.
Secondly, because there are much cheaper ways of stopping patent trolls from patenting obvious things: publish them, register them as prior art at the USPTO.
It's naive to think that the only way patents are used is to 'go after' projects. 99% of the time, patents of this sort are used in discrete discussions with potential clients. "You know, we hold a patent on that... (hint hint)". This is enough to scare the customer into at least not using a rival product, open source or not. Indeed, patents that make it to court tend to die rather faster than patents used under the table.
The irony of this patent is that technically, it's not that interesting. Dynamic message routing on XML is not difficult, but not efficient. It's much faster to pre-calculate routing keys and indices, as the existing AMQP exchanges do.
So I think Red Hat are simply playing the game of collecting software patents like points.
However, I really expected better from Red Hat.