New York University

Jonathan Chao

Yang Xu

Cing-Yu Chu

The problem of recovering from multiple link failures in a way that is quick, avoids loops, avoids packet modifications, and that avoids significant modifications to existing routers is solved by: (1) associating a network rerouting unit (NRU) with each of the plurality of routers; (2) configuring each router so that if a link (or more specifically, any link) directly connected to the router fails, the router redirects any packets whose next hop is a port terminating an end of the failed link to the NRU associated with the router; (3) executing a routing protocol on each of the NRUs whereby each NRU will learn a topology of the communications network; (4) receiving by a first NRU, a packet redirected from the router associated with the first NRU; and (5) responsive to receiving, by the first NRU, the packet redirected from the router associated with the first NRU, (i) identifying a link directly connected to the router as a failed link using a destination address in the redirected packet, and the topology of the network learned by the first NRU, (ii) determining an alternative path to the destination address of the redirected packet bypassing the identified failed link, and (iii) tunneling the redirected packet to an intermediate node on or associated with the alternative path using encapsulation. NRUs provide more programmability and better flexibility, thereby allowing network operators to deploy new network functions and features on demand in a timely and agile manner. Such NRUs provide resilience as a network function ('RaaNF') that can be plugged into a running network easily and help the network recover from link failures without modifying existing routers.

Growing community of inventors

Holmdel, NJ, United States of America

Jonathan Chao