Redundancy scheme for network processing systems

Abstract

A redundancy architecture is described for network processing systems which allows the network to recover from failure of a network processing system without interruption in service. The redundancy architecture allows network processing systems that use state information to associate network traffic into discrete flows, to provide system level redundancy to prevent service outages, or loss of network traffic resulting from a failure in any single network processing system. The redundancy architecture includes an out-of-band network link between the redundant network processing systems. The out-of-band network link allows the network processing systems to exchange state and other data as necessary. By maintaining the state data not only for the network traffic being processed by the network processing system, but also the state data for the network traffic being processed by its mate network processing system, either network processing system can assume the network traffic of its redundant mate in the event of a failure.