Scenario-based configuration management for flexible experimentation infrastructures

Configuration management is a key issue in flexible networking experimentation infrastructures (i.e., testbeds). However, manual procedures have several drawbacks (such as high time consumption in tedious and mechanic operations or the probability of introducing human errors). Moreover, inter-testbed scenario reutilization (i.e., the same network configuration replicated from one testbed to another) is difficult to achieve. In order to solve this limitations, automatic mechanisms are desirable in order to make testbeds profitable tools for their users. Our work describes one of such approaches, defining a model-driven methodology for automatic testbed reconfiguration. It is based in a high-level testbed-independent model (thus being applicable to any testbed in general) which is used to describe desired experimentation scenarios. Those scenarios are then particularized to testbed-specific models for automatic deployment and management by model-based tools. The paper also provides practical use cases applying the methodology in two different situations (VNUML-based virtual testbeds and the GMPLS-enabled optical networking ADRENALINE testbed^reg) thus assessing the feasibility and generality of our solution.