In a typical production environment, you set up the different API Manager components (API Publisher, Gateway, Store and Key ValidatorManager) in separate servers so that you can scale them independently. You also install multiple instances of a component in a cluster to ensure proper load balancing. When one node becomes unavailable or is experiencing high traffic, another node handles the requests.
...