Towards Software-defined Fog Computing via Named Data Networking

According to the fog computing paradigm, a plethora of Internet of Things (IoT) applications will benefit from the distributed execution of services and applications, processing and storage in proximity to data sources. However, new networking primitives and service orchestration mechanisms need to be designed to cope with the dynamic, distributed, and heterogeneous nature of the fog resources. A solution to such a challenging deployment would be the coupling of two revolutionary future Internet paradigms, Software-Defined Networking (SDN) and Named Data Networking (NDN). Indeed, on the one side, SDN can support wise orchestration decisions by leveraging a centralized intelligence and a global view of resources. On the other side, NDN well matches the service-centric nature of fog computing by letting (named) services be discovered regardless of the identity of the specific executor. In the paper, we dissect the strengths of this integrated solution, with a special focus devoted to evaluate the benefits of the NDN stateful data plane coupled with the centralized SDN control plane, when compared to a traditional IP-based host-centric software-defined approach.