Self-modeling based diagnosis of network services over programmable networks
Résumé
In this paper, we propose a multilayer self-diagnosis framework for network services within the software-defined networking and network functions virtualization environments. The framework encompasses 3 main contributions: (1) the definition of multilayered templates to identify the components to supervise across the physical, logical, virtual, and service layers. These templates are also finer-granular, extendable, and machine-readable; (2) a topology-aware and a service-aware self-modeling module that takes as input the templates, instantiates them, and generates an on-the-fly diagnosis model, which includes the physical, logical, and the virtual dependencies of network services; (3) a topology-aware and a service-aware root cause analysis approach that takes into account the network services views and their underlying network resources observations within the aforementioned layers to automate the diagnosis of programmable networks. We also present extensive simulations to prove and evaluate the following aspects: a fully automated diagnosis model generation and a fine-grained and reduced uncertainty diagnosis of the root cause for network services failures including those of their underlying resources. We include in this extended paper relevant state-of-the-art on topology and service aware diagnosis approaches for different types of network technologies, a deeper insight of our approach and problem formalization, and additional results