ACM ICN 2019, Hong Kong, CHINA
MENU

Half-day Tutorial: Service-based Routing at the Edge

Future scenarios, such as AR/VR, pose challenging latency and bandwidth requirements in 5G. This need is complemented by the adoption of cloud principles for providing services, particularly for virtualizing service components with which virtualized instances can appear rapidly at different execution points in the network. The solution seems simple in providing service endpoints close to the end user but early service break-out is currently limited to routing requests to Point-of-Presence (POP) nodes provided by few global CDN players deep in the customer network. In this tutorial, we present an approach that transforms the edge of the Internet into a rich service-based routing infrastructure with services being provided through edge compute nodes. We will provide insights into the architecture, its protocols, as well as implementation, evaluation and deployment insights.

Tutorial Overview

Trainers

  • Dirk Trossen

    InterDigital Europe, Ltd

    • Dirk Trossen is a Senior Principal Engineer at InterDigital Europe, the European branch of InterDigital Inc. His main responsibility lies in establishing the European presence of InterDigital through engagements within the EU-funded Horizon 2020 workprogramme as well as within UK-funded efforts. Dirk has more than 20 years of experience in network architectures, services and wireless technology. His main contributions can be found in the area of inter-domain networking as well as seamless handovers and new service concepts for operators. He has a long-standing experience in leading large-scale research consortia, most notably the European efforts PSIRP, PURSUIT, POINT and RIFE, currently leading the H2020 FLAME efforts. He has been performing ICN research since 2008, more recently with the goal of evolutionary introduction of ICN concepts in upcoming 5G networks. Prior to joining InterDigital, Dirk was co-founder of TecVis LP, a UK-based software solution company in the mobile, context-aware solution space and he held prior positions as a Senior Researcher with Cambridge University, Chief Researcher with BT Research and as a Senior Principal Scientist at Nokia Research. Until recently, he was a research affiliate with the Advanced Network Architecture group at MIT CSAIL. He holds a Ph.D. degree in Computer Science from Technical University of Aachen, Germany. He published more than 85 peer-reviewed papers in international conferences and journals and has currently more than 30 international patents.

       

  • Martin Reed, Mays Al-Naday

    Essex University

    • Mays AL-Naday is a Lecturer in the School of Computer Science and Electronic Engineering, University of Essex, UK. Prior to that, she worked as a senior research officer in the Network Convergence Laboratory (NCL), University of Essex. She worked as a DevOps Engineer in British Sky Broadcasting, and a research assistant in the. Her research focuses on future network architectures, including Information Centric Networking, IoT communications, Fog computing, next generation content delivery networks and security and Quality of Service in 5G and beyond architectures. She has been the organiser of the SIGCOMM workshop on Mobile Edge Communications (MECOMM), 2017 and 2018, and the IFIP NETWORKING workshop on Information Centric Fog Computing (ICFC). She has actively contributed to the ICN projects PURSUIT and POINT and currently exploiting outcomes of those project in developing secure Fog substrate within the ongoing H2020 project, SerIoT.

       

Motivation

The emergence of 5G systems facilitates many use cases; particularly those that rely on low latency communication as well as high bandwidth for delivery of, e.g., virtual reality content. Service execution is assumed to be highly flexible, driven by the adoption of cloud design principles which see services being provided potentially near end users. Examples for such use cases are, 5G mobile networks with service-based control plane architectures, those control planes being realized over a pure software-defined Layer 2 network that interconnects regional data centers. In addition, services for virtual reality, industrial as well as vehicular applications at the very edge of the network are being pursued in work on Multi-access Computing (MEC) or Fog computing, driven by the edge densification in terms of network and computing capabilities that is being outlined not only for 5G but also beyond.

Catering for these stringent requirements of low latency, high throughput and flexibility in service execution has become increasingly challenging since early service termination is limited to providing requests to well-managed POPs deep in the customer network, while providing services closer to end users would need to rely on service routing capabilities at the edge, where IP routing has not been established yet. Furthermore, traffic in present-day TCP/IP stack design needs to traverse several layers in a number of elements for the provisioning of an end-to-end service, while rigid service bindings hinder the ability to quickly redirect relations to nearer service execution endpoints. In other words, solutions are required that turn the current Layer 2 access network into an environment that can flexibly route (Internet) service requests.

In this tutorial, we provide insights into a solution that addresses these challenges by radically flattening the Internet protocol stack with each Internet service residing directly on top of a Name-based Routing (NbR) layer. Such direct mapping of HTTP and other services, including IP itself will allow for utilizing Layer 2 multicast forwarding capabilities, significantly reducing bandwidth requirements for multi-user HTTP-based services such as those employed for virtual reality. At the infrastructure level, we utilize the recent introduction of standardized programmable forwarding solutions, such as OpenFlow-based SDN switches. These technologies make changes in the network infrastructure a matter of updating the controlling software, as opposed to expensive hardware replacements. We use this foundation to realize an efficient path-based forwarding capability within an operator network, enabling extremely responsive and efficient name-based routing of Internet services within an autonomous domain, while retaining full compatibility with the existing Internet protocol interfaces towards peering networks and legacy devices. We show that all of this can be accomplished without per flow state in the network, with computational requirements similar to present-day networks.

We believe that ICN provides unique routing capabilities, such as inherent multicast support and content-oriented rather than location-oriented routing, that lend themselves to utilization in a service routing solution at the edge. By replacing the underlying IP routing with an ICN-based counterpart, while retaining the semantics of IP-based services, we will be able to gain from those capabilities while providing a backward-compatible introduction of ICN and improving on those existing IP-based services, hence increasing the incentive for said replacement of IP routing.

With the emergence of orchestration-based virtualization approaches, deployment of novel solutions in this space is enabled, complemented with a deployment in a network slices, as also recognized in the IRTF ICNRG deployment consideration work. As shown in own trial experiences, the time is right to propose such new routing approach, achieved through the orchestrated deployment of new service handling functions into the edge network.

Type of the Tutorial

The tutorial is lecture style with demonstrations in item 4 and 5 being provided through installations on the presentation laptop.

Outline of the Tutorial

1) Background for service routing at the edge (20 minutes)

Providing insight into mobile edge computing, 5G control and user plane architecture as well as driving use cases for the edge.

2) Network architecture & protocol operations (90 minutes)

Overview of network architecture and detailed insights into forwarding, and protocol mapping operations required.

3) Evaluation insights (60 minutes)

Overview of opportunities to test against as well as evaluation results.

4) Platform insights (45 minutes)

Overview of deployable platform, including orchestration framework.

5) Deployment (60 minutes)

Overview of existing deployments as well as deployment of an example service with running.

Requirements for Attendees

The solution being presented in the tutorial is based on the ICN variant developed in the European FP7-funded PURSUIT efforts with architecture and solution level differences in terms of name resolution and forwarding in the network. As background information, the references in [1] and [2] provide suitable background information on this ICN variant as well as the used path-based forwarding over an SDN-based transport network infrastructure. We recommend that those references are being studied by tutorial participants. Furthermore, the reference in [3] provides details of an early realization of the work with a focus on transferring IP packets over an ICN network, with newer work focusing on HTTP-based services being presented in the tutorial.

References

[1] D. Trossen, G. Parisis, Designing and Realizing An Information-Centric Internet, IEEE Communications Magazine Special Issue on “Information-centric Networking”, July 2012.

[2] Martin J. Reed, Mays Al-Naday, Nikolaos Thomos, Dirk Trossen, George Petropoulos, Spiros Spirou, “Stateless multicast switching in software defined networks”, In proceedings of ICC 2016, Kuala Lumpur, Maylaysia, 2016.

[3] G. Xylomenos, Y. Thomas, X. Vasilakos, M. Georgiades, A. Phinikarides, I. Doumanis, S. Porter, D. Trossen, S. Robitzsch, M. J. Reed, M. Al-Naday, G. Petropoulos, K. Katsaros, M. -E. Xezonaki and J. RiihijärviIP “Over ICN Goes Live,” Proceedings of EuCNC 2018.