Visual-INT : coleta, processamento e visualização granular de metadados INT para gerência SDN

• Ano de defesa: 2019
• Autor(a) principal: Zanetti, Alisson Borges
• Orientador(a): Trentin, Marco Antônio Sandini
• Tipo de documento: Dissertação
• Tipo de acesso: Acesso aberto
• Idioma: por
• Instituição de defesa: Universidade de Passo Fundo
• Programa de Pós-Graduação: Programa de Pós-Graduação em Computação Aplicada
• Departamento: Instituto de Ciências Exatas e Geociências – ICEG
• País: Brasil
• Palavras-chave em Português: Redes de computadores - Gerência; Telemetria; Processamento eletrônico de dados
• Área do conhecimento CNPq: CIENCIAS EXATAS E DA TERRA::CIENCIA DA COMPUTACAO
• Link de acesso: http://tede.upf.br:8080/jspui/handle/tede/1904

In computer networks, monitoring is an important tool to meet resource management. Together, the technological evolution and the growth of the networks make new possibilities for this type of task. Applied to recent technological environments like Software-defined Networking, In-band Network Telemetry aims to watch programmable devices by catching, encapsulating, and sending their metadata through the network flow. However, the use of data after its extraction is not defined by telemetry. Because of it, a lot of solutions addressed to this gap started emerging. Nevertheless, most of them are in simulated environments, do not have an immediate practical use, also bear on generic views from the collected data. To overcome that scenario, this work presents Visual-INT, a hardware-versed collector that captures, analyzes, stores, and response telemetry metadata. To do this, Visual-INT employs packet processing programming techniques, such as extended Berkeley Packet Filter and eXpress Data Path, in a Netronome card that catches telemetry traffic generated by other similar cards. Visual-INT promotes the understanding of the metadata now available on the cards, such as device latency and flow path, useful in monitoring delays and identifying points of failure as well as achieving higher hardware computation levels. The main propose of this work is to improve the telemetry gather techniques by offloading collector programs into SmartNIC, along with to offer a broad solution that is suitable for real environments.