Discussions

News: ActiveMQ 3.0 released

  1. ActiveMQ 3.0 released (26 messages)

    Codehaus has announced the release of ActiveMQ 3.0, an open-source messaging framework and JMS 1.1 provider.

    From the release notes, significant changes are:
    • a new 'discovery' transport protocol for a pure client-only peer network - a self discovering cluster which automatically finds the brokers available on the network (as opposed to the 'peer' protocol which is a complete peer based JMS network).
    • migrated from the org.codehaus.activemq to org.activemq package name hierarchy
    • improved support for more JDBC databases for persistence
    • an optimised wire protocol (which is unfortunately not compatible with 2.x)
    • more test cases and fixes of the JCA Container
    • various performance enhancements and bug fixes

    Threaded Messages (26)

  2. ActiveMQ 3.0 released[ Go to top ]

    I am excited about the goals of this project, but I do have concerns about the release strategy.

    How can you perform a major release of a message queueing system without addressing a known bug that causes messages to disappear?
  3. ActiveMQ 3.0 released[ Go to top ]

    How does this OS product compare to IBM MQ series ?? Especially the performance.
  4. ActiveMQ 3.0 released[ Go to top ]

    Look at the source and fix it yourself :-)
  5. ActiveMQ 3.0 released[ Go to top ]

    Did you completely miss the point or are you simply sarcastic ?
  6. ActiveMQ 3.0 released[ Go to top ]

    Somewhat sarcastic.
    Yes this is a glaring bug that should be fixed especially so because the reporter seems to have done most of the legwork (offering a very decent explanation of what might be going wrong)

    Sorry, I meant no offence
  7. ActiveMQ 3.0 released[ Go to top ]

    Did you completely miss the point or are you simply sarcastic ?

    He did have the smiley.
  8. Hi,

    If performance is what you are looking for, take a look at the open source project "MantaRay". it is a distributed, scalable and extremely fast messaging layer.

    www.mantamq.org

    hope i could help

    shai
  9. ActiveMQ 3.0 released[ Go to top ]

    I am excited about the goals of this project, but I do have concerns about the release strategy.How can you perform a major release of a message queueing system without addressing a known bug that causes messages to disappear?

    It does take us a reasonable amount of time to investigate customer & user test cases and scenarios to see if indeed there is a bug or some configuration issue involved etc. Whenever we do investige and find an issue we make sure we write a test case to ensure the issue remains fixed going forwards.

    We never release software when there is a known bug which we can reproduce, though we don't guarrentee to have closed every single reported incident in our issue tracker - particularly ones without detailed instructions on how to reproduce.

    FWIW this issue you highlighted, we still don't yet know how to reproduce (as things work great for us).

    James
    Protique
    Enterprise Open Source
  10. ActiveMQ 3.0 released[ Go to top ]

    James,

    sadly the entire release note is undermined by Corby's post. Being responsive to his post is the only way to claw it back.

    Jonathan
  11. ActiveMQ 3.0 released[ Go to top ]

    FWIW, I find ActiveMQ very useful for testing, mainly b/c it's pretty lightweight and flexible. I dont know if it's being used in production for any medium/large projects (as opposed to Tibco/MQSeries, etc), but I would be interested to find out.
  12. ActiveMQ 3.0 released[ Go to top ]

    FWIW, I find ActiveMQ very useful for testing, mainly b/c it's pretty lightweight and flexible. I dont know if it's being used in production for any medium/large projects (as opposed to Tibco/MQSeries, etc), but I would be interested to find out.

    Yes, ActiveMQ is being used today in production by a number of our customers and we're actively engaged right now in taking ActiveMQ into production for a number of other customers. If you mail me privately I can let you know some of them.

    James
    Protique
    Enterprise Open Source
  13. ActiveMQ 3.0 released[ Go to top ]

    My personal impressions on ActiveMQ:
    1) Unstable. The code (along with version numbers) changes very often to be propely tested.
    2) Lack of clustering/distributed messaging capabilities. Server does not support sending messages to remote queues, distributed durable subscribtions.
    3) JMS API speculations. Destinations are created on the fly and cannot be properly managed.
    4) Lack of security. No way to setup access rights to destination.
  14. ActiveMQ 3.0 released[ Go to top ]

    My personal impressions on ActiveMQ:1) Unstable. The code (along with version numbers) changes very often to be propely tested.
    We've just come off the back of a period of fairly intense development, you're right. Most of the functionality we need is there now; we're focussing mostly on stability going forward.

    Right now we've a fairly complete test suite (TCK, unit tests and load tests) though we're working with customers to add more scenarious and deployment configurations to our automated testing. With messaging there are so many different scenarious, topologies and requirements customers come up with - so our primary focus going forward is to ensure we automatically test these to ensure ActiveMQ is rock solid.

    2) Lack of clustering/distributed messaging capabilities. Server does not support sending messages to remote queues, distributed durable subscribtions.

    I don't follow? We support a federated network of servers with local and remote destinations (both store and forward and distributed destinations) and distributed subscriptions and message flows. Is there a specific topology you don't think we support?
    3) JMS API speculations. Destinations are created on the fly and cannot be properly managed.

    This is actually a feature lots of customers & users like, not an issue :). We can create destinations on the fly without long and laborious initial administation. e.g. compare getting MQSeries setup to something like Rendezvous. Once destinations exist you can manage them - but there's no absolute reason to have to manually configure them first. See the next issue for how we can lock down destinations so only specific users can use certain destinations...
    4) Lack of security. No way to setup access rights to destination.

    While not as advanced yet as some of the commercial alternative's we've a Security Plugin which can be used to implement fine grained security access to brokers and destinations.

    e.g. we've a provider which supports JASS to authenticate and JACC to authorize the user operations.

    If you have different security infrastructure or requirements, its easy to drop in other providers.

    I hope that helps give you a better idea of where we are today with ActiveMQ.

    James
    Protique
    Enterprise Open Source
  15. Native MQ[ Go to top ]

    Hi James,
    First, kudos to AMQ team for the great effort.
    I would like to know whether are you affiliated with the project to develop a native MQ product to be used especially in financial domain. As much as I know from public announcements, some big financial companies are developing such a product and will release it as open source.

    A comment: There are not any native, high-performance open source MQ product to be used in C/C++ applications. This market is occupied by Tibco, MQSeries,
    I think we (open sourcer) should develop a native MQ broker (maybe via porting AMQ) to replace the dust of legacy messaging systems.
  16. Native MQ[ Go to top ]

    Hi James,First, kudos to AMQ team for the great effort.
    Thanks Cetin
    I would like to know whether are you affiliated with the project to develop a native MQ product to be used especially in financial domain.
    Yes we are. I can't say much more though due to NDAs I'm afraid
    As much as I know from public announcements, some big financial companies are developing such a product and will release it as open source.A comment: There are not any native, high-performance open source MQ product to be used in C/C++ applications. This market is occupied by Tibco, MQSeries,I think we (open sourcer) should develop a native MQ broker (maybe via porting AMQ) to replace the dust of legacy messaging systems.

    Yes. Details of the project has already been leaked and I can't really say much more other than ActiveMQ interoperates fully with the AMQ C/C++ project. I'll be able to say more when the NDA is lifted

    James
    Protique
    Enterprise Open Source
  17. Native MQ[ Go to top ]

    Details of the project has already been leaked and I can't really say much more other than ActiveMQ interoperates fully with the AMQ C/C++ project. I'll be able to say more when the NDA is lifted

    I'm terribly interested in this aswell.
    Do you have any idea when the project will become public (and open source if I remember correctly)?
    Thank you.
  18. Native MQ[ Go to top ]

    Details of the project has already been leaked and I can't really say much more other than ActiveMQ interoperates fully with the AMQ C/C++ project. I'll be able to say more when the NDA is lifted
    I'm terribly interested in this aswell.Do you have any idea when the project will become public (and open source if I remember correctly)?Thank you.

    I'm afraid not. Its in the hands of a large financial organisation and their legal department to decide when the time is right.

    James
    Protique
    Enterprise Open Source
  19. ActiveMQ 3.0 released[ Go to top ]

    Does anyone know how ActiveMQ would compare with OpenJMS? I've been using OpenJMS for a while now and have been pretty happy but it doesn't seem like it's being actively developed, and it doesn't seem to play too nice with other J2EE servers. Has anyone worked with both, and can anyone compare them both feature- and (more importantly) performance/stability- wise?

    Drew
  20. ActiveMQ 3.0 released[ Go to top ]

    Does anyone know how ActiveMQ would compare with OpenJMS? I've been using OpenJMS for a while now and have been pretty happy but it doesn't seem like it's being actively developed, and it doesn't seem to play too nice with other J2EE servers. Has anyone worked with both, and can anyone compare them both feature- and (more importantly) performance/stability- wise?Drew

    I"m obviously biased, but here's a few differences

    * ActiveMQ fully supports JMS 1.1 and works in any J2EE 1.4 container (Geronimo, JBoss 4, WL 9 etc)

    * ActiveMQ is much faster (we'll publish some stats soon, but for now you can try them yourself with our JMeter benchmark suite

    * ActiveMQ supports auto-discovery, auto-reconnection, clustering, embedded brokers, networks of brokers and many more transports and topologies (like a pure peer network, federated networks with store and forward, pure in VM messaging for unit testing etc).

    I'm sure there's many more; those are the ones off the top of my head.

    James
    Protique
    Enterprise Open Source
  21. unscientific results[ Go to top ]

    when I was developing the JMS sampler for JMeter, I did some tests with OpenJMS, Orion JMS, ActiveMQ and Joram. My rough tests showed ActiveMQ is quite a bit faster than OpenJMS. I also ended up discovering a bug with Orion's JMS client implementation. If I find the number I'll post them. Not sure if I deleted them or not.
  22. How are folks using JMS?[ Go to top ]

    How are you folks using JMS in your applications?

    Did your applications start off from a design perspective around an anynchronous model, or did you graft it on later in select places?

    How many queues/topics et al are you using?

    I'd just like to see some case studies of discussions of the style of apps that rely on this kind of technology, etc.

    It seems popular, it seems straight forward, maybe I'm "thinking too much". I have some thoughts of my own, but I'm just curious how others are using it. Has anyone gone overboard, "let's make everything aysnchronous" and all of a sudden you have a bunch of Message Beans vs Session Beans and then pulled back to the "happy medium"?

    Just like to see some other designs and such of asynchronous systems in the wild.
  23. How are folks using JMS?[ Go to top ]

    Just like to see some other designs and such of asynchronous systems in the wild.

    I am currently working on an open source "Enterprise Service Bus" framework, which is basically a distributed and highly-decoupled integration framework.
    Information is available at: http://www.infonatural.com/esb/
    And the Sourceforge project-page is at: http://sourceforge.net/projects/infonatural-esb/

    There is a "concepts"-document available if you want to read up on the thinking behind it. It is still an early draft but should give a good overview of the thinking.

    This particular framework is still very much in its early-days, lots of cleaning-up and some rework to be done.

    But I have based it quite heavily on several years of EAI experience working as a contractor for companies such as IBM, Sun Microsystems and Accenture (both best and worst practices have influenced me in my thinking).
    The concepts in themselves borrow heavily from David Chappell's "Enterprise Service Bus", and Gregor Hohpe's and Bobby Woolf's "Enterprise Integration Patterns".
  24. How are folks using JMS?[ Go to top ]

    We're using JMS for a transaction processing system.
    The transactions are set off by the user. The transaction data goes through a very fast formal check and is stored. The user is notified of acceptance. This is all non-asynchronous.
    The data is then sent to a queue (pending) and a separate program processes it through a complex system involving remote servers and such. The processing results are put in a second queue (processed). The first process gets notified of this and updates the main database and then notifies the user of the outcome.
    So: push(data) -> store(db,queue:pending) | <- onMessage(queue:pending) -> doProcess(complex) -> store(queue:processed) | <- onMessage(queue:processed) -> store(db) -> notify(user)
    where | is a system's inflection point, an asynchronous link.

    The system was designed from the start with this capability in order to scale. We're using JMS over Oracle AQ.
  25. activemq vs joram[ Go to top ]

    so it seems the flavours of the month for opensource jsm are joram from objectweb and activemq.

    can anyone compare?
  26. activemq vs joram[ Go to top ]

    so it seems the flavours of the month for opensource jsm are joram from objectweb and activemq.can anyone compare?

    One more vote for a comparison article. :)

    In general, I wouldn't mind a bit more JMS and integration articles on TSS..
  27. More JMS on TSS[ Go to top ]

    But wouldn't messaging and integration articles be a little bit too "enterprise" for TSS?

    ;-)