Titulo Estágio
Exploring P4 & KDN to improve service performance
Áreas de especialidade
Comunicações, Serviços e Infraestruturas
Sistemas Inteligentes
Local do Estágio
LCT
Enquadramento
In recent years, there has been a focus shift to improve the performance, flexibility, and reliability of communication infrastructures. For users, latency (lag) has become a critical aspect of the performance of applications such as online video gaming and video streaming. By default, IP networks offer a best-effort service, meaning that routers will do their best to deliver data packets without offering guarantees. To optimize the performance and thus the user experience, techniques related to traffic engineering, load balancing, Quality of Service (QoS), among others are being implemented in the network devices.
Programming Protocol-independent Packet Processors (P4) is an open-source domain-specific programming language for network devices, specifying how data plane devices (switches, routers, NICs, filters) process packets (https://opennetworking.org), allowing the implementation of customized forwarding logic in programmable networks. The P4 ecosystem includes an extensive range of products, projects, and services.
This work focuses on using P4 to create custom forwarding techniques that improve the Quality of Service for applications taking advantage of a global view of the infrastructure using Knowledge Defined Networking (KDN), which incorporates a Knowledge Plane (KP) in Software-Defined Networking (SDN) architecture. Machine Learning (ML) techniques are applied in the KP to generate knowledge from the network information and from past executions to generate knowledge that can guide the forwarding tasks implemented with P4.
The work comprises the interaction of an international team, being supervised by researchers from the Laboratory of Communications and Telematics (LCT) in the Centre for Informatics and Systems of the University of Coimbra (CISUC), Portugal, and from the Institute of Computing (IC) from the University of Campinas (UNICAMP), Brazil.
Objetivo
The work consists of designing and developing forwarding solutions using P4 and KDN to improve the performance of applications.
Plano de Trabalhos - Semestre 1
(First Semester)
Phase 1: Study of the state-of-the-art on programmable networks
Phase 2: Review of forwarding techniques and KDN paradigm
Phase 3: Familiarization with P4 and KDN
Phase 4: Service requirement and network resource definition
Phase 5: Prepare midterm defense document
Plano de Trabalhos - Semestre 2
(Second Semester)
Phase 7: Implement a baseline forwarding mechanism using P4 and KDN
Phase 8: Implement a second forwarding mechanism using P4 KDN
Phase 9: Validate the mechanisms and analyze their results
Phase 10: Prepare the dissertation document
Condições
The work will be performed in the LCT Laboratory (Departamento de Engenharia Informática, Universidade de Coimbra, Portugal), in close collaboration with researchers from the Institute of Computing (University of Campinas, Brazil).
Observações
There is the possibility of awarding the internship with a scholarship, according to the candidate’s profile.
---
This work will be co-supervised by Prof. Edmundo Madeira from the University of Campinas (UNICAMP), Brazil.
Orientador
David Abreu
dabreu@dei.uc.pt 📩