The Architecture for ActiveMQ 6 is designed to be flexible and
extensible. Its being built on an OSGI kernel - http://servicemix.apache.org/SMX4KNL/index.html
.
With the input of SonicMQ architects - we are building out the
enterprise class features of ActiveMQ, but we will be accommodating
more wire formats - including AMQP -as well as true restful API.
On 8 Mar 2009, at 04:59, bwtaylor wrote:
There is more noise again around my shop regarding AMQP. The "AMPQ
== RedHat"
assertion doesn't play because you've got RabbitMQ and Qpid. Yes, QPid
started as a Red Hat code drop, but everybody understands that
Apache is
robust to domination by any one vendor, a fact that ApacheMQ and
Camel both
demonstrate well. But even if what you say is true, if Red Hat is
the only
vendor that comes forward with a solution for them, that's not a
good place
for you to be. Nobody fears Red Hat lock in.
I do not need AMQP per se: what I need is high quality cross platform
messaging. So if you've changed your plans and aren't going to
tackle AMQP
because it isn't simple to implement for existing broker platforms,
why not
team up with the folks you mention and come up with something that
is. I
expect that the reason the AMQP spec writers didn't come up with a
solution
that could be bolted on to existing brokers is because they got the
cold
shoulder from the projects you listed.
STOMP is not the answer. It's too simplistic and asking ruby and
python apps
to confine their messaging capabilities to what STOMP provides is
met with
the same enthusiasm you'd get asking java shops to give up JMS for
it. The
stomp python clients all have various states of disrepair. The ruby
one
works, but there's critical unresolved bugs related to activemq's
stomp
implementation anyhow: AMQ-2137, AMQ-1941, AMQ-1873, AMQ-1807. Also
stomp
won't have keep alive until v1.1 (AMQ-2019). We've seen this leak
sockets to
the point where we hit the ulimit max and our broker hangs.
rajdavies wrote:
The AMQP reality is that only new message brokers will implement it -
simply because you'd have to re-write the message broker to
accommodate it. Which is why you won't see any of the traditional
messaging platforms like Webshpere MQ, SonicMQ or Tibco EMS, RV
implementing any time soon. We would love to offer full support for
it
in ActiveMQ - but that's going to take lot of investment and a lot
of work.
Its a shame the AMQP spec writers didn't concentrate on making AMQP
simple to use and implement for existing messaging platforms in the
same way STOMP did - which is why both OpenMQ and RabbitMQ support
STOMP - and SonicMQ will probably being doing the same in the future
too.
The AMQP protocol is open argument kinda disappears up its own
backside once folks realize the cost of entry - that a vendor has to
start from scratch to implement it - so in reality AMQP == RedHat
currently for enterprises. Ironic - when the whole point of AMQP was
to try break vendor lock-in!
cheers,
Rob
Rob Davies
http://fusesource.com
http://rajdavies.blogspot.com/
On 26 Jan 2009, at 18:49, bwtaylor wrote:
I'll also express strong interest in AMQP and I'll take the liberty
of saying
that most people using stomp for cross platform integration with
ActiveMQ
should be expressing interest. With the influx of enterprise apps
being
written in dynamic languages, AMQP offers high end messaging
features in a
platform agnostic way.
I would also caution against assuming that the people who want AMQP
for
messaging are likely to seek you out to express that interest. If
I'm a ruby
on rails or a django shop and I figure out I need a messaging
solution for
cross platform integration, I'll soon have an interest in AMQP.
When I look
for implementations I'll find RabbitMQ or Redhat Messaging, or
AMQP in
Fedora 10 and never think about ActiveMQ.
In fact, if you don't support AMQP that will be a talking point
against
deploying ActiveMQ in an IT environment where ruby or python apps
exist.
I've already had that happen at my company and I've played down AMQP
as
still in development, not quite fully baked, but now with Fedora 10
touting
AMQP as a major new feature, that argument's lifespan is ending and
people
are becoming more aware of it.
James.Strachan wrote:
2008/12/19 loctorp <boris.kartasch...@logica.com>:
Hi everyone,
I was wondering about the current status of AMQP implementation
into
acticeMQ. On the project page it states, that there is a sandbox
version
and
that developement has been paused.
As we are interested in using activeMQ together with AMQP we were
wondering
if this status has changed and/or are interested in the up-to-date
outlook.
The status hasn't changed since that wiki page was written.
Welcome -
you're the first person ever to express any interest in AMQP with
ActiveMQ :)
--
James
-------
http://macstrac.blogspot.com/
Open Source Integration
http://fusesource.com/
--
View this message in context:
http://www.nabble.com/actual-status-of-activeMQ-and-AMQP-tp21092034p21671180.html
Sent from the ActiveMQ - User mailing list archive at Nabble.com.
--
View this message in context:
http://www.nabble.com/actual-status-of-activeMQ-and-AMQP-tp21092034p22395001.html
Sent from the ActiveMQ - User mailing list archive at Nabble.com.
