Java EE 8 High Performance

Fallback code structure

When you were just handling a failover across servers, it was not very complicated, as you would probably have the list of hosts in the Client class, and iterating over them was almost natural. Now that we iterate over different implementations, it is less natural and we need an orchestrator bean. Concretely, for the previous example, which first calls a remote service and then falls back on a local hardcoded instantiation, we would need the following:

• A remote client
• A local mock implementation
• A facade (orchestration) service, which is used everywhere, so we can leverage this failover logic

When you start integrating lots of services, it is not very convenient.