<div class='slidealt'>Experience kvm <a title='virtualization for embedded heterogeneous arm core platforms' href='/en/products'>virtualization extensions</a></div> <div class='slidealt'>Benefit from custom <a title='kvm on arm services full virtualization' href='/en/services'>virtualization services</a></div> <div class='slidealt'>KVM on ARMv7 and ARMv8 <a title='kvm-on-arm open source smu extensions' href='/en/solutions/guides/vfio-on-arm/'>IOMMU full virtualization</a></div> <div class='slidealt'>Virtualization research projects <a title='ARM multicore kvm open source' href='/en/research'>in cloud and embedded systems</a></div> <div class='slidealt'>Virtualization solutions for heterogeneous <a title='ARMv7-ARMv8 virtualization open source solutions' href='/en/solutions'>ARM multicore systems</a></div>

Virtual Open Systems Scientific Publications

Cross-compartment Virtio-loopback: a bare-metal virtualization solution for the edge

Cross-compartment Virtio-loopback - Edge virtualization solution - conference ESARS-ITEC 2024

Event

The 2024 International Conference on Electrical Systems for Aircraft, Railway, Ship Propulsion and Road Vehicles (ESARS) and International Transportation Electrification Conference (ESARS-ITEC 2024).

Keywords

Edge computing, Virtualization, Mixed-critical systems, Virtio-loopback.

Authors

Anna Panagopoulou, Alvise Rigo, Daniel Raho (Virtual Open Systems SAS).

Abstract

With the evolution of 5G technologies and beyond, paradigms like the Mobile Edge Computing (MEC) mandate the deployment of services at the edge of the network and close to the users. Contributing to further reducing the latency and improving network performance, these paradigms are essential constituents of a networked infrastructure supporting ultra-reliable and low-latency services, as in the case of the recently emerging intelligent transportation services (ITS). With MEC being coupled with Network Functions Virtualization (NFV), these services are deployed upon the cellular V2X interface as Virtualized Network Functions (VNFs) across the various network domains, including the network edge. This paper proposes a virtualization solution suited for VNF deployments of these services at mixed-critical, embedded edge servers. The solution suggests that VNFs interfacing abstracted Virtio devices are hosted at individual and isolated bare-metal Operating Systems (OSes) or compartments, co-executing at the same hardware with different criticality. The current proof-of-concept targets ARMv8 embedded servers and introduces a split client-server Virtio design, extending an existing virtualization framework across bare-metal compartments. Finally, benchmarks on a virtualized embedded board show that the Cross-compartment Virtio-loopback design does not introduce extra overhead compared to the reference implementation.

Introduction

The reinforcement of vehicular networks with intelligent transportation systems (ITS) has opened the way towards improving traffic management, enhancing road safety and minimizing the impact of vehicles on the environment. In these regards, the role of modern interfaces that connect vehicles to everything (V2X) is indispensable, as they enable real-time communication and data exchange between vehicles and the broader infrastructure.

In the context of the cellular V2X interface, recent mobile network trends such as the Multi-access Edge Computing (MEC) place significant emphasis on the edge domain, advocating for the deployment of transportation services at edge servers, in close-proximity to vehicles. Thanks to these trends, the computational capacities of powerful edge devices can be now more than ever exploited to the greatest extent while aiding to achieve the real-time, low-latency requirements of the ITS services. At the same time, the network services are deployed in the infrastructure and across all network domains as chains of Virtualized Network Functions (VNFs). Being decoupled from proprietary hardware, these virtualized functions have facilitated a cloud-to-edge continuum in terms of deployment.

In this regard, the question arises of whether uniformly applying a state-of-the-art virtualization solution across the network, though beneficial for consistency, truly optimizes performance. In light of this, we introduce the Cross-compartment Virtio-loopback solution which aims to re-shape the virtualization environment at the edge, making it particularly suitable for reliable and low-latency services deployment, while maintaining the cloud-to-edge uniformity at a high degree.

Specifically, the Cross-compartment Virtio-loopback is a lightweight, bare-metal virtualization solution, aimed for mixed-critical embedded architectures, such as the ARMv8 and the RISCv. The inherent capability of these architectures to accommodate multiple critical levels, led to the concept of compartments, which are Operating Systems with different critical levels, able to co-execute isolated in the same embedded hardware. At its core, the Cross-compartment Virtio-loopback aims to reinforce the edge devices with the ability to host the virtualized functions at individual compartments, deployed in a bare-metal fashion on the physical hardware. Proposed as an alternative approach to employing virtualization at the network edge, the solution seems promising due to ensuring hardware-enforced isolation and bare-metal execution latency for the deployed compartments.

Cross compartment virtio loopback presentation

Cross compartment virtio loopback presentation

Access the full content of this publication

Login or register to access full information