As discussed in the section called “Location transparency”, partitioned objects are also distributed objects. This provides application transparent access to the current active node for a partition. Applications simply create objects, read and modify object fields, and invoke methods. The TIBCO BusinessEvents® Extreme runtime ensures that the action occurs on the current active node for the partition associated with the object.
When an active node fails, and the partition is migrated to a new active node, the failover to the new active node is transparent to the application. No creates, updates, or method invocations are lost during partition failover as long as the node that initiated the transaction was not the failing node. Failover processing is done in a single transaction to ensure that it is atomic. See Figure 7.7, “Partition failover handling”.
