Advanced topics

In the following sections, we will discuss some advanced topics in regards to clustering. We will especially see how you can rely on a TCP stack to create your cluster configuration, to use the TCP protocol for communication and transfer purposes, that is, the cluster member's advertising and state replication. Additionally, we will see what clustering means for applications bundling Enterprise Java Beans and how clients are affected.

For simplicity, we will stack with standalone mode, using the standalone-ha.xml configuration file.

Clustering using TCP

As we have seen at the beginning of the chapter, to achieve cluster capabilities, JBoss relies on two components: JGroups and Infinispan.

The first one is used for communication purposes; the second one is used for persistence purposes. To switch from a UDP cluster to a TCP cluster, we need to modify the JGroups configuration.

Let's first create a standalone clone folder, named tcp-cluster-1, as follows:

cd $JBOSS_HOME
cp -a standalone...