FAULT-TOLERANT NETWORKING

摘要

Methods and apparatus for implementation of fault-tolerant networks provide a network fault-tolerance manager for detecting failures and manipulating a node to communicate with an active channel. Failure detection incorporates one or more methods, including message pair and link pulse detection. Failure recovery includes switching all node data communications to a stand-by channel or switching just those nodes detecting a failure. Communication between nodes provides the distributed detection, with detecting nodes reporting failures to their network fault-tolerance manager and the network fault-tolerance manager broadcasting the failure recovery to all nodes. The network fault-tolerance manager is middleware, residing with each node in a logical hierarchy above a physical layer of a network and below a transport layer of the network. The approach is particularly suited to Ethernet LANs and capable of using commercial off-the-shelf network components.