Is it worth adapting sub-wavelength switching control plane to traffic variations?

This paper proposes a novel Metropolitan Area Network (MAN) architecture called Multi-hEad sub-wavElength swiTching (MEET). Compared with the current architectures, MEET proposes to aggregate traffic using passive optical nodes instead of using electrical nodes (switches and routers). Several options regarding a potential control plane are compared in terms of resource allocation efficiency. Two options are relative to the burst assembly process whereas another option is relative to the dynamicity of the resource allocation process. Performance evaluation is carried out using a simulation platform fed by real traffic traces captured on a French operator's metropolitan network. The QoS delivered to three different classes of service has been assessed in terms of latency and jitter. Obtained results show that a control plane that does not adapt to short-term variations of the real traffic may provide QoS levels compatible with an operational MAN.