A single point of failure
Even if the different Java EE layers of an application are clustered, there are still other parts of the infrastructure that may be single points of failures. Examples are network hardware (switches and so on), load balancers, and database systems. These need to be investigated further to minimize the impact of a failure.
Another important consideration during risk analysis is to divide the available cluster nodes between several data centers when possible. This ensures that the application will run even if one of the data centers is unavailable.
