Microsoft patents method to pinpoint where network failures occur

1. A method, implemented at a computer system in a network that includes one or more processors, for localizing network failure in the network with a topology, the topology of the network indicating a plurality of nodes in the network and how the plurality of nodes are connected to each other, the method comprising: enumerating, by the one or more processors, at least one path between the computer system and a path destination node in the network through one or more intermediary nodes in the network; enumerating, by the one or more processors, a plurality of sub-paths between the computer system and the path destination node, each sub-path having one or more intermediary nodes of the at least one path as a corresponding sub-path destination node; for each of the plurality of sub-paths, generating a first number of sub-path data packets, each of the first number of sub-path data packets configured to traverse a round trip of the corresponding sub-path from the computer system to the corresponding sub-path destination node, and then from the corresponding sub-path destination node back to the computer system; sending, by the computer system, the first number of sub-path data packets over the network; and receiving, by the computer system, a second number of at least a portion of the sub-path data packets; determining a probability that a data packet can successfully traverse the round trip of the corresponding sub-path by dividing the second number by the first number; and identifying an intermediary node where network failures have occurred, based on a plurality of probabilities that a data packet can successfully traverse the respective round trips of the plurality of sub-paths.