Analysis and optimization of fast failure detection and failover for HA cluster-based systems
Yang Liu, Jie-Bin Xu · 2010
With the rapid development of network technology and the increasing expansion of its size, there are more and more services need to be provided by network, so modern users have an increasing demand for making servers that operate and provide these services highly available. Obviously, stand-alone system is unappeasable, we must use the cluster as the server that operate those services especially the important ones. At present, the major work-styles to improve HA (High Availability) are Hot Standby and Dual Active which can basically satisfy the demands of SMEs. But for the others like large servers what could provide many crucial services like Mail Service and database service, neither can well satisfy the demands, because a single minor failure may lead to many major losses in this case. A fast failure detection and failover HA system has been proposed, the system here is multi-active. The scheme that will design could find an optimum member to take over the broken-down services in the shortest time.