3 Replies Latest reply on Jun 29, 2011 4:47 AM by ronsen

    difference between weblogic cluster and jboss cluster

    wutongjoe

      Hi all,

       

      after reading some documents about jboss,I  have some  questions for clustering .

       

      1)  It seems that unlike weblogic having an admin server controlling all managed server nodes  inside/outside cluster,jboss instances in a cluster are all equally running and communicating through multicast,there is no master server and slave servers ,am I right ?

       

      2) is there any way to manage the behavior of ALL jboss nodes in a cluster ?  just like weblogic did ? start all servers,stop all servers,monitor health status of  any jboss instance through a central point gui ? suppose I had a cluster with 100 nodes distributed on 50 physical machines with 50 different IP

       

       

       

      thanks in advance

        • 1. difference between weblogic cluster and jboss cluster
          wdfink

          Hi again Josph

           

          1) I'm happy with JBoss clustering using MCast, so you are able to add a node without administrating a other one.

          It is possible to use TCP cluster configuration if you are not able to use multicast, I never configure it but I'm sure it will work

          The Master/Slave will come automaticaly, one of the server's will become the master (it will change if this server goes down)

           

           

          2)

          This can be done by JBossOn an additional administration server for administration and monitoring

          • 2. difference between weblogic cluster and jboss cluster
            wutongjoe

            Hi Wolf-Dieter Fink , thanks again for your helpful information.

            • 3. Re: difference between weblogic cluster and jboss cluster
              ronsen

              Just a remark, with the master/slave thing i discovered that the masternode has at least 20-40 more threads open and doing some stuff compared to the slave nodes. Another thing i discovered was the failover was in avg. a bit slower failing over from slave to master instead from master to slave. It can be that it was a statistical inaccuracy but it was a bit strange. Didnt know it affects it measureable.