Joint fault tolerant and latency-aware design of multilayer optical networks

When designing a multilayer IP-over-WDM optical network, an important requirement is finding the most cost-effective design that is tolerant of a set of selected failures arbitrarily defined, so the 100% of the IP traffic survives for them. A second concern is the need to guarantee a maximum end-to-end latency for IP traffic (e.g. 50 ms in continental US). Unfortunately, fault tolerance and latency at the IP layer are performance merits extremely challenging to enforce in multilayer networks, since they depend on both the IP routing over the lightpaths and the lightpath routing over the fibers, and how they change during recovery processes. There is little research considering one of these concerns separately, and to the best of the authors' knowledge, just preliminary works on how to design the network jointly considering both requirements. Note that a joint optimization is needed, since e.g. backup routings that do not guarantee end-to-end latency may be useless. In this paper, we address this problem, and present a multilayer planning algorithm to dimension IP/OSPF-over-WDM networks with three main recovery schemes: IP-only restoration, 1+1 optical protection followed by IP restoration, and optical-followed-by-IP restoration. Results are provided to validate the algorithm, and feed a discussion on the joint fault tolerant and latency-aware multilayer network design under those schemes.

I. Introduction

In the context of IP-over-WDM multilayer backbone networks, IP and WDM layers are often operated independently, by separate departments. On the one hand, IP departments optimize the routing over the topology of IP links, consisting of end-to-end lightpaths. Each lightpath is seen by the IP layer as a pipe, a direct connection between two routers. On the other hand, the optical layer is operated by different personnel, in charge of providing the requested lightpaths: (i) deciding on its route through the physical topology of fibers, and (ii) managing the transponder banks, optical add/drop multiplexers and the rest of optical equipment.