US Patent 7797572 Computer system management method, management server, computer system, and program

This invention provides a method of controlling switching of computers according to a cause of failure without preparing one standby node for each active node. For n active nodes (200), m standby nodes (300) of different characteristics (in terms of CPU performance, I/O performance, communication performance, and the like) are prepared. The m standby nodes (300) are assigned in advance with priority levels to be failover targets for each cause of failure. When a failure occurs in one active node (200), a standby node that can remove the cause of the failure is chosen out of the m standby nodes (300) to take over data processing.