A cluster is the binding of two or more Virtual Hosts that reside on separate nodes, such that each Virtual Host runs servlets across all of the nodes and processes sessions on any one of those nodes. The implementation of clustering in WebSphere Application Server allows for failover, which preserves session data integrity and the common pool of sessions in the event of a system failure in one or more of the clustered Java virtual machines (JVMs) running servlets within a WebSphere Application Server Virtual Host group. It also allows for load-balancing, whereby the session workload is distributed among the Virtual Hosts comprising the cluster.

WebSphere Application Server maintains session clusters using a database. In a clustered environment, the session may be accessed on any one of the Virtual Hosts in a cluster; which one is actually accessed will be transparent to the end user. During the processing of a session transaction, if the Virtual Host fails for any reason during the WebSphere HttpSession transaction then the update to the database does not occur, but the common pool of sessions continues to function (including the session being processed during the failure, minus any updates made during the failed transaction). For non-catastrophic failures (for example, when the virtual host remains functional), any changes made to the session during the transaction attempt are rolled back and the session reverts to its state prior to the start of the transaction. Otherwise, once the transaction has completed and the changes are committed, the session is still accessible regardless of the failure of an individual node.