In-Network IdentifiCation and chAining: um mecanismo de identificação de tráfego e encadeamento no plano de dados

• Ano de defesa: 2021
• Autor(a) principal: Matos, Guilherme Mendes Vieira de
• Orientador(a): Verdi, Fábio Luciano
• Tipo de documento: Dissertação
• Tipo de acesso: Acesso aberto
• Idioma: por
• Instituição de defesa: Universidade Federal de São Carlos; Câmpus Sorocaba
• Programa de Pós-Graduação: Programa de Pós-Graduação em Ciência da Computação - PPGCC-So
• Palavras-chave em Português: 5G; P4; Plano de dados; Dispositivos programáveis
• Palavras-chave em Inglês: Network slice; Dataplane; Service function chaining; Programmable devices
• Área do conhecimento CNPq: CIENCIAS EXATAS E DA TERRA::CIENCIA DA COMPUTACAO::METODOLOGIA E TECNICAS DA COMPUTACAO
• Link de acesso: https://repositorio.ufscar.br/handle/20.500.14289/15424

Inside the 5G Core (5GC) a myriad of (virtual) functions may be deployed so that different treatment can be given for traffic coming in and out to/from the Radio Access Network (RAN). One of these key functions is Service Function Chaining (SFC) supported by SRv6 state-of-the-art protocol. In this work, we present an SFC P4-based solution for traffic identification and chaining using SRv6. We call this function In-Network Classification and chAining (INCA), which is deployed entirely in the data plane using a Netronome Agilo SmartNIC. The INCA is deployed just before the User Plane Function (UPF) inside the 5GC and is capable of observing the traffic coming from and going to the RAN so that it classifies and creates the proper sequence of services to be followed by every specific flow. Our results show that INCA performs the task of packet classification and chaining perfectly with a minimal Flow Completion Time (FCT) impact when compared to the same environment without it.