The Execution of Event-Action Rules on Programmable Network Processors

This paper evaluates the ability of programmable network processors (NPs) to perform application-specific processing that is structured as sets of interdependent eventaction rules sharing joint state. Our intent is twofold: (1) to assess the ability of NPs to deal with the complex application-specific code and state maintained by eventaction rules like those used in business processes also termed (termed business rules), and (2) to create system solutions that permit developers to dynamically map such application-level service code to appropriate sites in target distributed platforms comprised of hosts, network processors, and embedded server systems. A specific outcome of our work presented in this paper is the creation of a simple, efficient dynamically reconfigurable rule engine for a network processor, able to execute rules at the Gigabit speeds required by the network links attached to it. Business rules like those found in the Operational Information Systems used by companies like Delta Air Lines are used to demonstrate rule engine capabilities and overheads. A second outcome of our work is its demonstration of the flexibility and ease of reconfiguration associated with a network processor-resident rule engine, where rules can be added and removed whenever appropriate (i.e., hotswapping) without compromising the processor’s ability toLarge-Scale Network Simulation: How Big? How Fast? maintain high performance for its ongoing processing