Protected Packet Routing: Achieving Fast Failure Recovery in the IP Layer

Failure recovery in IP networks has gained much attention from both academic and industrial circles. The main challenge of this issue is how to achieve fast recovery while keep the implementation complexity and resource requirement at low level. Current IP networks rely on either routing table recalculation or lower layer protection for failure recovery. Although resource-efficient, routing recalculation takes seconds even minutes to complete. On the other hand, lower layer protection requires much redundant resource to achieves the speed of millisecond recovery. This paper proposes a pure IP layer scheme to achieve fast failure recovery with neither redundant resource reservation nor routing recalculation on the fly, called protected packet routing (PPR). The idea is to include a protection output port in each entry of the routing tables. Once a failure occurs, a number of routers switch to their protection ports for packet forwarding, thus bypass the failed equipment automatically. The challenges of this approach are: 1. Is a single protection port always sufficient to guarantee failure recovery? 2. How to find such ports? 3. How to coordinate the routers for failure recovery without little signaling? 4. How to realize distributed implementation? Focusing on single-link and single-node failure scenarios, this paper provides answers to the above questions and presents the performance evaluation of the proposed algorithms. Based on this work, it is promising to study double-failure scenario such that the survivability of IP networks can be intensified to tolerate the majority of network failures.