Transparent Dual-Node Hot Standby Scheme Based on PCIe Synchronization

The core tasks in the computing and control systems have stronger demands for better usability and safety. With the constant emerging of redundancy and exception handling methods, the consistency of state, RAM, I/O, abnormal event detection, and primary /secondary transparent handover methods are proposed, and a model of a task transparent dual-node hot standby system based on PCIe instant syncing is constructed. For the sake of speed of RAM state syncing, a I/O-dominant RAM syncing method is designed based on the PCIe NT interconnection channel, that is, the I/O-dominant method is used for checkpoint syncing to reduce no-required RAM cache transferring process between the primary and secondary servers and reduce the time of server state syncing. The proposed transparent fault-tolerant technology can greatly reduce the additional overhead introduced by the dual-node hot standby method. And the cost of the dual-node hot standby method is only very small in terms of performance. A quickly failover can be obtained from the primary the secondary. The performance analysis shows our proposal ensures the node performance and the continuous operation of vital tasks.