On Scenario-based Model-driven Configuration Management for Flexible Networking Experimentation Infrastructures

Abstract

The goal of the present dissertation is to improve the current state of the art in configuration management in networking experimentation infrastructures (testbeds). In particular, our goal is to solve the scenario technology-dependency problem, so the same scenarios can be reused in different testbeds. In order to achieve this goal, three objectives have to be addressed. First, to develop a model-driven configuration management architecture that decouples scenario specifications from the testbeds in which they will be finally deployed. This has to be done without modifying existing testbeds or their management tools. Secondly, a Testbed Independent Model (TIM), based on DMTF's CIM, used as main architectural component in order to provide scenario technology-independency. Finally, a generic and systematic methodology to transform TIM scenarios to the specification formats used in the different current or future testbeds. In order to achieve these objectives, our work includes an extensive analysis of the most salient experimentation infrastructures existing today, paying special attention to their configuration management systems. In addition, different technologies to model and transform management information are analyzed, in order to find the best one to base our architecture on. This dissertation also includes an experimental validation to assess the feasibility of our architecture, which is applied to two different testbeds used for real world experimentation. Additionally, several application cases are described, both for novel testbed architectures (such as the ones defined in Future Internet initiatives) and production environments.