Uma Abordagem de Fatiamento de Rede entre Múltiplos Sistemas Autônomos

Abstract

Resource sharing is common within operating systems and computational hardware infrastructures. The scientific community, driven by flexibility and scalability requirements, has expanded the concept of sharing resources to the scope of communication network infrastructures. For this, enabling technologies such as Software-Defined Networking, Network Functions Virtualization, Segment Routing, and cloud computing paradigms leverage new approaches to sharing resources. Sharing network resources by organizing them functionally and logically in the network connectivity ecosystem can be recognized as a network slice. Network slicing has received joint efforts from academia and industry because of new technological enablers such as programmability and virtualization. Network slicing has been extensively explored in the specification and standardization process for new mobile network architectures, specifically \textit{5G}. Therefore, the proposals in the literature focus predominantly on providing network slicing in the context of mobile networks. Approaches that proposed network slicing beyond mobile networks proved to be functionally ineffective in providing network slicing across multiple Autonomous Systems (ASs). Thus, this thesis builds and evaluates a network slicing method between multiple ASs based on Internet routing algorithms. This thesis answers the assumed hypothesis by adding dynamism to the network slice process by enabling the user or administrator to perform network slice specifies definitions, such as service parameters and the path along with Internet routers to the network slice. Besides, this thesis brings novelty with a conceptual and technological framework for performing recursive slicing. Experiments measuring the applicability, relevance, and performance of the network slicing approach were conducted and contrasted with traditional methods. Experimental results in significant scenarios suggest that the method of this thesis functionally and in performance surpasses traditional state-of-the-art methods. Furthermore, experiments demonstrate that dynamism in choosing the network slice path can improve the quality of experience for real-time applications.