Advanced search
Start date

VNF Benchmarking and tuning and SFC performance prediction

Grant number: 21/06909-7
Support type:Regular Research Grants
Duration: January 01, 2022 - December 31, 2023
Field of knowledge:Physical Sciences and Mathematics - Computer Science - Computer Systems
Cooperation agreement: MCTI/MC
Principal researcher:Judith Kelner
Grantee:Judith Kelner
Home Institution: Centro de Informática (CIn). Universidade Federal de Pernambuco (UFPE). Ministério da Educação (Brasil). Recife , SP, Brazil
Assoc. researchers: ASSIS TIAGO DE OLIVEIRA FILHO ; Djamel Fawzi Hadj Sadok ; Eduardo James Pereira Souto ; Eduardo Luzeiro Feitosa


The complexity of the TCP/IP stack has increased with the introduction of virtual machines, containers, hypervisors, virtual network drivers, programmable network cards, software-defined networking technology, and multi-core CPU architectures. The new reality requires the development of a framework that guides VNF (Virtual Network Function) designers in the selection of operating system components, network drivers, virtualization software, CPU usage pinning, TCP/IP stack bypass and offloading mechanisms, intelligent and programmable network cards, communication libraries and their correct configuration to optimize the performance of a VNF. The framework will support dynamic adjustment of the number of allocated resources, including memory and CPU cores, depending on load, to ensure that a VNF meets your service specifications and that the user experience is not degraded. It also includes a VNF performance prediction model and benchmarking that gives VNF designers insights into the performance of a VNF in different deployment scenarios. The model required a relationship between allocated resources and the resulting service performance. The forecast model simulates a "what-if" scenario. Although there are many solutions for benchmarking VNF performance, it is not always easy to accurately predict the performance of a service chain (Service Function Chain) made up of NFVs from different manufacturers, from the performance benchmark Isolated VNF. Therefore, a more efficient and streamlined approach is needed to ensure that an SFC meets performance specifications while requiring fewer design iterations. The research will develop smarter, faster, and more advanced models to estimate the KPIs (key performance indicators) of SFCs. (AU)

Articles published in Agência FAPESP Newsletter about the research grant:
Articles published in other media outlets (0 total):
More itemsLess items

Please report errors in scientific publications list by writing to: