TY - GEN
T1 - P4NFV
T2 - 14th International Conference on Network and Service Management, CNSM 2018 and Workshops, 1st International Workshop on High-Precision Networks Operations and Control, HiPNet 2018 and 1st Workshop on Segment Routing and Service Function Chaining, SR+SFC 2018
AU - He, Mu
AU - Basta, Arsany
AU - Blenk, Andreas
AU - Deric, Nemanja
AU - Kellerer, Wolfgang
N1 - Publisher Copyright:
© 2018 IFIP.
PY - 2018/12/20
Y1 - 2018/12/20
N2 - Current architecture proposals for Network Function Virtualization (NFV) do not integrate hardware-accelerated network function implementations. Recent research studies have shown that pure software-based implementations cannot achieve the needed line rates for todays network services. We propose P4NFV to fill this gap. Making use of an additional abstraction layer, P4NFV is an architecture that can achieve software-based network function implementations as well as handle P4 for programming protocol-independent packet processors. With P4NFV, network operators can still instantiate network functions that are specified in terms of computing and storage hardware, while making use of the performance improvements of P4-enhanced networking hardware. Moreover, in order to take the fast changing nature of todays network services into account, P4NFV integrates mechanisms to reconfigure P4-based network functions at runtime: another missing gap in literature. Based on a proof-of-concept implementation of P4NFV for four network functions, we show promising measurement results. Whereas the network function implementations tailored towards reconfigurations add only marginal overhead, even configuring network functions at runtime does not notably affect network service operations with higher latency or severe packet loss.
AB - Current architecture proposals for Network Function Virtualization (NFV) do not integrate hardware-accelerated network function implementations. Recent research studies have shown that pure software-based implementations cannot achieve the needed line rates for todays network services. We propose P4NFV to fill this gap. Making use of an additional abstraction layer, P4NFV is an architecture that can achieve software-based network function implementations as well as handle P4 for programming protocol-independent packet processors. With P4NFV, network operators can still instantiate network functions that are specified in terms of computing and storage hardware, while making use of the performance improvements of P4-enhanced networking hardware. Moreover, in order to take the fast changing nature of todays network services into account, P4NFV integrates mechanisms to reconfigure P4-based network functions at runtime: another missing gap in literature. Based on a proof-of-concept implementation of P4NFV for four network functions, we show promising measurement results. Whereas the network function implementations tailored towards reconfigurations add only marginal overhead, even configuring network functions at runtime does not notably affect network service operations with higher latency or severe packet loss.
KW - Network Architecture
KW - Network Function Adaptation
KW - Network Function Virtualization
KW - Programmable Data Plane
UR - http://www.scopus.com/inward/record.url?scp=85060922421&partnerID=8YFLogxK
M3 - Conference contribution
AN - SCOPUS:85060922421
T3 - 14th International Conference on Network and Service Management, CNSM 2018 and Workshops, 1st International Workshop on High-Precision Networks Operations and Control, HiPNet 2018 and 1st Workshop on Segment Routing and Service Function Chaining, SR+SFC 2018
SP - 90
EP - 98
BT - 14th International Conference on Network and Service Management, CNSM 2018 and Workshops, 1st International Workshop on High-Precision Networks Operations and Control, HiPNet 2018 and 1st Workshop on Segment Routing and Service Function Chaining, SR+SFC 2018
A2 - Salsano, Stefano
A2 - Riggio, Roberto
A2 - dos Santos, Carlos Raniery Paula
A2 - Ahmed, Toufik
A2 - Samak, Taghrid
PB - Institute of Electrical and Electronics Engineers Inc.
Y2 - 5 November 2018 through 9 November 2018
ER -