Fine-grained Composition of Service Function Chains in the Programmable Data Plane

9^th February 2023, 16:00
Posco Tso
Loughborough University

Abstract

Dynamic service function chains (SFC) are enabled by network function virtualization on general purpose servers. The emergence of programmable data plane (PDP) has offered a new way for the deployment of SFC. However, the implementation of network functions is constrained by resource limitations in PDP (e.g., compute and memory resource). Moreover, most of existing works do not consider the optimization of state information (e.g., registers), which is essential for stateful network functions. In this talk, I will introduce pSFC -- a fine-grained SFC deployment scheme in the PDP for tackling this challenge. We first model network functions as control flow graphs (CFG) and the process of deployment as a one big switch (OBS) problem, and then propose an Integer Linear Programming (ILP) model for resource optimization for the OBS problem, which is NP-hard. To solve this problem efficiently, pSFC first composes multiple SFCs for eliminating redundant resources, decomposes the compound CFG based on the resource limitation per stage, and finally maps OBS into the substrate network. We have implemented pSFC in both bmv2 software switch and an Intel Tofino P4 hardware switch. Evaluation shows that pSFC reduces switch costs 45.7% and average latency by 22% without compromising throughput.