Uma arquitetura de rede programável para redes orientadas à informação com replicação de conteúdo em nuvens privadas

• Ano de defesa: 2018
• Autor(a) principal: Nascimento, Erick Barros
• Orientador(a): Macedo, Douglas Dyllon Jeronimo de
• Tipo de documento: Dissertação
• Tipo de acesso: Acesso aberto
• Idioma: por
• Programa de Pós-Graduação: Pós-Graduação em Ciência da Computação
• Palavras-chave em Português: Computação em nuvem; ICN; SDN; Sistemas distribuídos; Redes programáveis; Cache; Proxy reverso; Controlador
• Palavras-chave em Inglês: Cloud computing; Distributed systems; Programmable networks; Reverse proxy; Controller
• Área do conhecimento CNPq: CIENCIAS EXATAS E DA TERRA::CIENCIA DA COMPUTACAO
• Link de acesso: http://ri.ufs.br/jspui/handle/riufs/10684

Internet leads with the information, services and content are made available in digital format. Access to such content can be possible from anywhere, by the most varied devices. However, the configurability of the conventional network architecture is hampered by the diversity of embedded systems deployed in the network assets, where the information that use its resources does not have an effective management, while specific problems are not easily detected by the systems and infrastructure teams, for instance, duplicated HTTP traffic requests in the network segment. This research conducts an experimental study with proposal of a decentralized architecture for content networks. The architecture centralizes all management by decoupling the concentrators control of the network, allowing the application of the configurations on the fly. With the advent of Software Defined Networking (SDN), new possibilities for network management are emerging for new contexts for the Information-centric Network (ICN). These possibilities can be achieved through network programming with remote extension of the data plane to the cloud. This work presents and validates a prototype of content networks, made available in an established domain and replicated between them, where each endpoint perform a microcache to optimized the request and response time. In addition, a minimized copy of this backend is placed on the cloud in a Virtual Private Server (VPS)) enabling high availability of the environment. System has two reverse proxies capable of taken control of the data plane without perception to the user, keeping high performance and optimization through the Least_Conn load balancing algorithm. Validation took place through three scenarios. In the first, it was evaluated aspects inherent to the operation of a computer network, therefore, latency, packets loss and flow rate. In the second scenario, aspects were evaluated regarding the tunneling for the cloud with directional and bidirectional request to obtaining delay and packets loss. Finally, aspects of replication, caching, contention band, optimization, reachable bandwidth, Cache Hit Ratio (CHR), Mean Time Between Failure (MTBF) and Mean Time To Repair (MTTR) were presented, achieving good results with ICN and SDN. The work finalize pointing to study of controller latency due to variations in the system, but this did not impact the research objectives.