2 Replies Latest reply on Nov 4, 2002 11:41 AM by Sacha Labourey

    Dynamic Clustering

    Jaksa Vuckovic Newbie

      Hi!
      I hope this is the right forum for this kind of question.
      While reading the JBoss 3.0 Clustering documentation I found a mention that a future version of JBoss will include dynamic clustering.
      I take part in a quite large accademic project involving the universities of Madrid, Bologna, Zurich, McGill, Trieste and Newcastle. The project is about (among other things) extending JBoss to enable adaptivity to non-functional user requirements. In other words a client should be able to specify something like on how many instances shall its request be processed.
      This involves dynamic configuration of subclusters. I wanted to know where can we find more information about these future features, who is responsable of coordinating the clustering developement and examine a form of collaboration.
      I apologize again if this is not the right forum for a question like this.
      Best regards,

      Dr. Jaksa Vuckovic

        • 1. Re: Dynamic Clustering
          Sacha Labourey Master

          Hello,

          Some code for dynamic sub-division of a cluster in sub-clusters is already in JBoss but... it has been put in "stand-by" when Bill did the detached-invoker implementation. And yes, the goal is to re-plug that feature in, but nothing has been planned yet.

          Cheers,


          sacha

          • 2. Re: Dynamic Clustering
            Sacha Labourey Master

            Hello,

            Some code for dynamic sub-division of a cluster in sub-clusters is already in JBoss but... it has been put in "stand-by" when Bill did the detached-invoker implementation. And yes, the goal is to re-plug that feature in, but nothing has been planned yet.

            If you want to take a look at it, it is in the cluster module, somewhere in org.jboss.ha.hasessionstate.server (I did that originally for SFSB). You could indicate the sub-cluster minimal-size (e.g 2) and when one node died, the cluster was recreating sub-clusters by minimizing the new pairs (i.e. to avoid the need for a state transfer)

            Cheers,


            sacha