In this paper, we propose a P4-based Data Plane Programmability and Exposure framework (DPPx), which allows to enhance Network Function Virtualization (NFV) services. Network functions can be offloaded from VMs to the virtualized network infrastructure. We present a general approach, system architecture and use cases enabled by DPPx. The DPPx framework has been preliminarily evaluated in a testbed environment as a prototype implementation based on OpenStack Neutron, P4 language and BMv2 software switch.
1 February 2021 in IEEE Transactions on Network and Service Management
A hypervisor switch, such as Open vSwitch (OVS), plays a key role in the virtualized data centers implementing overlay networking to provide network isolation. The hypervisor switch is running on each compute node as a software application and switches packets between virtual machines. Software switches frequently require upgrading and customization of network protocol’s stack to introduce novel or domain-specific networking techniques. However, it is still difficult to extend OVS to support new network features as it requires mastery of network protocol design, programming expertise, and familiarity with the complex codebase of OVS. Moreover, there is currently no solution that enables the deployment of network features recompilation of OVS. In this paper, we present P4rt-OVS, an original extension of OVS that enables runtime programming of protocol-independent and stateful packet processing pipelines. It extends the forwarding model of OVS with the userspace Berkeley Packet Filter (uBPF), bringing a new extensibility mechanism. Moreover, P4rt-OVS comes with a P4-to-uBPF compiler, which allows developers to write data plane programs in the high-level P4 language, compile them to the BPF bytecode and inject them to the OVS packet processing pipeline at runtime. Our design results in a hybrid approach that provides P4 programmability without sacrificing the well-known features of OVS. We provide thorough performance results, including end-to-end performance tests, microbenchmarks and a performance test of an exemplary network application (Broadband Network Gateway). The performance evaluation shows that P4rt-OVS does not introduce significant processing overhead, yet enables runtime protocol extensions and stateful packet processing. Moreover, we discuss features of P4rt-OVS and provide programming guidelines to help developers achieve the best performance of P4 programs for P4rt-OVS.
In this paper we describe the IIP System designed, implemented and currently tested in the frame of the Polish national project entitled "Future Internet Engineering"1 (in Polish "Inzynieria Internetu Przysz"- IIP). The IIP system uses virtualized network infrastructure that allows to set a number of, named, Parallel Internets essentially differing in solutions for the data and control planes. More precisely, we design three Parallel Internets: (1) IPv6 QoS that basically explores Diff Serv and NGN architectures and uses TCP/IP, (2) Content Aware Network (CAN) specially designed for efficient content delivery, and (3) Data Streams Switching (DSS) for handling constant rate traffic with strong QoS requirements. For the IIP System we design a novel network infrastructure, which is based on the virtualization of the network elements (links and nodes). Such approach is on the line with the expectations of the network infrastructure for the Future Internet. In the paper we mainly focus on the system architecture, Parallel Internets and some aspects of network virtualization.