Universal Messaging 10.5 | Concepts | Performance, Scalability and Resilience | Failover
 
Failover
Universal Messaging clients, whether server to client or server to server, automatically handle disconnect and reconnection to any number of alternative Universal Messaging Realm servers.
If a Universal Messaging server becomes unreachable the Universal Messaging session will automatically try to connect to the next defined Universal Messaging server in the list of RNAME values provided when the nSession object was created. Universal Messaging clients can get this list of RNAMES from the server during the session handshake or programmatically at any time. The ability to pass RNAME values from server to client dynamically makes it very easy to manage client failover centrally.
Single Universal Messaging Realm
In a single server scenario upon receipt of an abnormal disconnection a Universal Messaging client will automatically try to reconnect. The back-off period and interfaces handling reconnection are all fully exposed in a public API. They can be used to trigger off any number of specific events.
Multiple Universal Messaging Realms
In a multiple server scenario Universal Messaging clustering (see Clusters: An Overview) allows for multiple Realms to act as live replicas of each other. This has huge benefits in terms of high availability.
Universal Messaging is also fully compatible with high availability and business contingency products implemented within the underlying operating system. This means that Universal Messaging is compatible with existing HA/BCP policies as well as providing its own in built fail over functionality.