An open GMPLS-enabled control plane testbed for remote development and experimentation of PCE-based path computation algorithms

The deployment of a distributed control plane allows optical transport networks to support the new emerging requirements such as real-time and flexible provisioning, multiple levels of QoS, and required network survivability, through providing a common set of functions and interconnection mechanisms defined under the GMPLS architecture. One of the main challenges in GMPLS-controlled transport networks is to develop efficient algorithms for dynamic path computation. For large networks, the path computation process may become computationally complex, requiring dedicated computational resources. To this end, the IETF has introduced a stateless Path Computation Element (PCE) entity in the GMPLS-enabled control plane, which provides advanced path computations serving requests from network nodes. The decoupling of the path computation function from the GMPLS-enabled control plane in a dedicated entity allows network operators to have more flexibility in the control of their networks, without being subject to the constraints of the network equipment vendors. In this context, there is a clear need for experimental performance evaluation of dynamic PCE-based path computation algorithms in GMPLS-enabled transport networks to obtain meaningful live concrete key performance indicators. However the vast majority of experimental platforms are conceived as a proof-of-concepts and only used for validation purposes. This paper presents an open GMPLS-enabled control plane testbed, based on the infrastructure of the ADRENALINE testbed, to allow researchers the remote development and experimentation of dynamic PCE-based path computation algorithms through an open API.