When two or more WildFly instances are started with an HA configuration (such as
standalone-ha.xml), they form a cluster. In each of these instances, the selected web applications and EJBs can be easily clustered. We will look at these later on in this chapter. First, however, we will have investigate the foundations of clustering in WildFly.
The low-level parts of cluster communications in WildFly, such as discovery and group handling, are handled by technologies from an open source project called JGroups (http://www.jgroups.org). Higher-level functions, such as replication and data handling are performed using Infinispan (http://infinispan.org).
The following text taken from the JGroups website provides us with ...