High availability architecture for network devices

Abstract

A method and apparatus allows for continued operation of one or more applications running at a network device with reduced delay despite crashes or failures at that device. The network device includes two or more supervisor cards for running the applications and a plurality of line cards. According to the invention, one supervisor card is designated the active supervisor card and one supervisor card is designated the standby supervisor card. As changes in state and other operating conditions take place on the active supervisor events are generating for passing at least some of this information to the standby supervisor where it is stored. Following a crash or failure of the active supervisor card, the standby becomes the newly active supervisor card. The standby supervisor performs a consistency check with the line cards and resets those that fail the check. The standby supervisor also determines which data records and state information stored at the standby supervisor are valid, and begins running the applications loaded onto the device. Those data records and state information determined by the standby supervisor to be valid are utilized by the applications in continuing their operation, while invalid data records and state information are discarded.