On the performance evaluation of distributed dynamic routing in GMPLS optical networks

This paper investigates the efficiency of deploying distributed dynamic routing using the GMPLS control plane in optical transport networks. Our main objective is not to propose new protocols or algorithms, but rather to find the limitations of the existing GMPLS distributed dynamic routing implementation and quantify its value-added proposition. Thus, we could provide guidance to service providers to minimize the cost and maximize the savings. Some of our findings through NSFNET simulations reveal that a) for distributed dynamic routing to be effective traffic loads should be less than 0.8 Erlang/per-wavelength, the number of wavelengths in a bundled traffic engineering (TE) links should be less than 10. b) Distributed dynamic routing can save at least 15% wavelengths per link and increase the traffic loads by 20%