Minimizing protection switching time in transport networks with shared mesh protection

Abstract

Summary Shared mesh protection (SMP) protects service traffic on a working path and requires coordinated use of the shared resources on a protection path when the traffic is routed through the protection path. If there are many intermediate nodes on the protection path, the delay caused by the coordination at each node increases the protection switching time, resulting in losing more traffic. Considering the capacity of a link and the bandwidth required by the protection paths using that link, it appears that coordination is necessary. However, looking closely at the relationships with other links and their protection paths, there exist resources that may require no coordination at all. This observation leads to the fact that protection switching time can decrease when bypassing some intermediate nodes for coordination. This fact introduces an optimization problem, in which the objective is to exclude these resources as much as possible from the set of resources for which coordination is required. A sub‐problem is finding these excludable candidate resources. For both the main problem and the sub‐problem, which are NP‐hard, we propose a greedy algorithm and a branch‐and‐bound algorithm, respectively. The simulations show that the results using the proposed algorithms are near‐optimal. Furthermore, in an actual topology configured with the SMP environment, the protection switching time is improved by excluding resources for which coordination is not required, compared to the existing Optical Transport Network (OTN) SMP mechanism.

Trade-off between Resource Efficiency and Fast Protection for Shared Mesh Protection

“…10) modeled on a real American network. The reason for adopting this topology in the experiment is that existing studies related to SMP ( [26], [29]) performed experiments on this topology. For verification of realistic scenarios, we assumed that there could be data-center cities for full-mesh connections.…”

“…For verification of realistic scenarios, we assumed that there could be data-center cities for full-mesh connections. To reflect the control overhead in the experiment, we assumed that Tα and Tβ are approximately 4.9 and 2 ms (referring to [26]) for the simulation experiments described in Table 3. Note that these parameter values can vary depending on the system and the implementation.…”

“…Through experimentation, the following two parameters were found to be most sensitive: the sharing rate related to the capacity of the link and the number of failures that can occur probabilistically. We defined a sharing rate [26], which indicates the degree to which the protection paths share the capacity of a link. The sharing rate for the links we set is expressed as a percentage, and the smaller this value, the more the resources are shared.…”

“…The maximum segment elimination problem (MSEP) is used to find transparent segments that do not require coordination operations, even if they are shared for multiple protection paths. MSEP has been studied mainly by maximizing the number of transparent segments [26]. The centralized software-defined networking (SDN)-based approach, considered to be the most preferable option for network control and management tasks, can pre-calculate protection paths and apply them to switches [27].…”

“…Referring to [26], we assume that a service succeeds in protection after Nf attempts to preempt the resources of the protection path. If the NT option is used, the tail-end node waits until it receives the NRNA message and then restarts the activation of the protection path from the tail-end node.…”