A Disaggregated Packet Processing Architecture for Network Function Virtualization

Abstract

Network Function Virtualization (NFV) promises to reduce the capital and operational expenditure for network operators by moving packet processing from purpose-built hardware to software running on commodity servers. However, the state-of-the-art in NFV is merely replacing monolithic hardware with monolithic Virtual Network Functions (VNFs), i.e. , software that realizes different network functions. This is a good first step towards transitioning to NFV, however, common functionality is repeatedly implemented in monolithic VNFs. Repeated execution of such redundant functionality is particularly common when VNFs are chained to realize Service Function Chains (SFCs) and results in wasted infrastructure resources. This stresses the need for re-architecting the NFV ecosystem, through modular VNF design and flexible service composition. From this perspective, we propose MicroNF ( $\mu $ NF in short), a disaggregated packet processing architecture facilitating the deployment of VNFs and SFCs using reusable, loosely-coupled, and independently deployable components. We have implemented the proposed system, including the different architecture components and optimizations for improving packet processing throughput and latency. Extensive experiments on a testbed demonstrate that: (i) compared to monolithic VNF based SFCs, those composed of $\mu $ NFs achieve the same packet processing throughput while using less CPU cycles per packet on average; and (ii) $\mu $ NF-based SFCs can sustain the same packet processing throughput as those based on state-of-the-art run-to-completion VNF architecture while using lesser number of CPU cores.