Sun Java System Message Queue 4.1 provides a highly available messaging service through a new "Highly Available" cluster type. In an MQ cluster of this type, all the broker instances would share a peer– to–peer relationship and all its broker instances would share a common persistent data store thus offering 'Data Availability.' Instances would automatically be able to detect if an instance fails and perform a takeover of the failed broker's persistent messages, dynamically through a takeover election. Application components that are deployed in the application server could thus leverage these availability features.
With this cluster type, there would not be any loss of transacted persistent messages to a queue or a durable topic subscription. Non-persistent messages or persistent messages to non-durable subscribers are likely to be lost when the broker that the client runtime is connected to is unavailable.