Transitioning between redundant computer systems on a network

Abstract

A method is provided for transitioning between redundant computer systems on a network. According to one aspect of the present invention, fault tolerance may be achieved in a network which includes one or more redundant computer systems by monitoring the status of a primary system and having a redundant system take over for the primary system upon determining the primary system has entered a predetermined state (e.g., lost power, CPU has gone down, etc.). Each host machine on the network includes an address translation table (e.g., an Address Resolution Protocol (ARP) cache) for purposes of dynamically mapping a logical address (e.g., an Internet Protocol (IP) address) to a physical address (e.g., a MAC address). Therefore, transitioning from the primary system to a redundant system that shares the same logical address may be accomplished by simply transmitting an ARP reply containing therein the shared logical address and the physical address of the redundant system or an ARP proxy server with which the redundant system is registered. In this manner, the host machines will update their address translation tables such that the shared logical address will be associated with the physical address provided in the ARP reply and further transmissions addressed to the shared logical address will in effect be redirected to the redundant system either directly or indirectly through the ARP proxy server.