RTGov has two mechanisms for processing events:
(a) Event Processor Network - a network of nodes where each node has an optional predicate to act as a guard, and a processor (java, mvel, drools) which may create further information that is then passed to other nodes/networks that have registered interest. Events sent between nodes/networks are distributed via jms, allowing load balancing across clustered servers. Events are also processed in a batch to minimise the transaction/communication overhead per event processed - successfully processed events in the batch are passed to the next nodes/networks, and unsuccessful events are retried.
(b) Event Validator - reuses the predicate/processor concept to validate events as they are reported within an embedded collector - these policies have the capability to block the business transaction that is reporting the activity.
It would be good to discuss how (a) could be handled by the new Alert module. I believe (b) could possibly be handled as an extended capability of an embedded “Agent”.
