P4SC: A High Performance and Flexible Framework for Service Function Chain

Abstract

Network function virtualization allows network functions to be implemented and managed flexibly as a service function chain (SFC) in the data plane to process flows. However, software-based SFCs lead to poor performance compared to proprietary middleboxes. Moreover, existing solutions tackling performance issues suffer from the development complexity incurred by hardware details. To address these problems, we leverage both the high performance of P4-capable devices and the high flexibility of P4 language. In this paper, we present a P4 Service Chaining framework (P4SC), which tackles multiple challenges for P4 to support the SFC implementation. P4SC provides a suite of primitives allowing efficient SFC expression, and a converter and a generator converting input SFC requests to the corresponding P4 program. Here, an algorithm based on longest common subsequence (LCS) is used to allow simultaneously implementing multiple SFCs. Moreover, P4SC offers a runtime manager for flexible SFC management at runtime. It also provides an automatic integration mechanism to integrate P4-based NFs into P4SC. We implement a P4SC prototype, which supports three types of P4-capable devices. The experimental results show that P4SC outperforms state of the arts with orders-of-magnitude SFC performance improvement while maintains high flexibility.

A Survey on In-Network Computing: Programmable Data Plane and Technology Specific Applications

“…The studies in [170], [171] design a framework for service function chaining (SFCs) on the P4-capable devices with the hardware capability as well as P4 programmability, to enhance performance, as well as flexibility in the implementation of Service Function Chains (SFCs). As shown in Fig.…”

“…There are studies that propose framework/deployment solutions to exploit in-network computing in NFV environment. Some studies focus on deployment of VNFs in network elements [164], [165], [167], [168], [169], [170], [171].…”

“…The main achieved outperforming in these criteria have been reported in Tables XII, XIII, XIV, and XV. For the purpose of better comparison, Table XVI compares the in-network implemented functionalities versus serve/dedicated hardware-based schemes, which are baselines for comparison in many studies e.g., [157], [169], [170], [171], [172]. We provide more details as below:…”

“…Latency reduction by 10× in comparison with a software-based cloud load balancer [140], up to 15× latency reduction in mobile applications due to FPGA acceleration at edge [144], latency less than 6 ms for end-to-end message delivery through mesh relay/routing nodes playing the role of fog nodes [58], as well as latency of 50 µs for in-network LTE protocol stack and gateways processing [147], and up to 25% reduction in handover time between Distributed Units in next generation RAN [149], have been reported in the literature. As examples of throughput enhancement, the evaluations on real world service function chains in [170], [171] illustrate that software-based NFV solution will have roughly 0.01 Mbps throughput in software-based approach, while it will end to roughly 10 2 Mbps throughput utilizing the in-network implementation of network functions. As another example, up to 102× throughput increcrement has been reported through offloading LTE EPC control plane operations to the programmable switches [150].…”

CHIMA: a Framework for Network Services Deployment and Performance Assurance

“…Other approaches for the composition of P4 functions have been studied in Hyper4 [8], which defines a P4 program that is able to emulate other P4 programs provided at runtime by the control plane and in P4Visor [9], which suggests a technique to merge multiple P4 programs into a single one, while retaining the functionality of all of them. An alternative take on the acceleration of SFCs with programmable data planes is explored with P4sc [10] and in [11], both of which consider the implementation of whole function chains on single switches, lacking the possibility of combining components of different technologies.…”