| Internet-Draft | SARNET-21 | July 2021 |
| Galis & Lou | Expires 27 January 2022 | [Page] |
This document provides an overview of the "Semantic Addressing and Routing for Future Networks" workshop (SARNET-21), which took place on June 10, 2021, in Paris, France and online as part of the IEEE International Conference on High-Performance Switching and Routing. The main goal of the SARNET-21 workshop was to explore, together with the research community, the use cases and network requirements in the domain of semantic addressing and routing, and identify potential research challenges to be tackled in the future.¶
Note that this document is a report on the proceedings of the SARNET-21 workshop compiled by the authors. It captures the views and positions of the workshop participants as expressed during the workshop.¶
This Internet-Draft is submitted in full conformance with the provisions of BCP 78 and BCP 79.¶
Internet-Drafts are working documents of the Internet Engineering Task Force (IETF). Note that other groups may also distribute working documents as Internet-Drafts. The list of current Internet-Drafts is at https://datatracker.ietf.org/drafts/current/.¶
Internet-Drafts are draft documents valid for a maximum of six months and may be updated, replaced, or obsoleted by other documents at any time. It is inappropriate to use Internet-Drafts as reference material or to cite them other than as "work in progress."¶
This Internet-Draft will expire on 27 January 2022.¶
Copyright (c) 2021 IETF Trust and the persons identified as the document authors. All rights reserved.¶
This document is subject to BCP 78 and the IETF Trust's Legal Provisions Relating to IETF Documents (https://trustee.ietf.org/license-info) in effect on the date of publication of this document. Please review these documents carefully, as they describe your rights and restrictions with respect to this document. Code Components extracted from this document must include Simplified BSD License text as described in Section 4.e of the Trust Legal Provisions and are provided without warranty as described in the Simplified BSD License.¶
Various networks, such as IoT networks, industry networks, data-center networks, content distribution networks, satellite networks, etc., have been emerging for the past decade in the light of digital transformation. These heterogeneous networks often adopt different architectures, topologies, as well as addressing and routing mechanisms. The satellite network is a typical example of a highly dynamic topology in which mobile network nodes can cause stability issues to existing routing protocols. IoT and manufacturing networks often adopt alternative addressing semantics that go beyond the network location defined in IPv4 and/or IPv6.¶
In simple terms, semantic addressing is about taking a regular address and assigning special meaning to some or all of the bits. Examples include multicast addresses, segment identifiers in segment routing, and network programming abstractions. Although the specific semantics help to facilitate addressing and routing within the network, scalability challenges arise from providing the definition of those semantics, and the interconnection among such networks and the Internet.¶
Several techniques have been proposed that modify/improve the default IP forwarding behaviors (such as least-cost path) to better meet the application requirements, based on additional information available in the packet (both in header and payload) and configurable policies in routers. Collectively, these mechanisms are sometimes referred to as "Semantic Routing" [I-D.king-irtf-challenges-in-routing]. The alternative semantics used to make routing decisions, together with the IP addresses, could be applied to a network overlay [RFC7665], or can be directly embedded into the address field, which is the case in some limited domains [RFC8799] such as LoRaWAN deployments [LoRaWan].¶
However, those ad-hoc solutions have been developed in a fragmented way, which creates interoperability issues between limited domains or between individual routers, and can lead to increased fragility or even security/privacy leakage. A more holistic approach can design the architectural patterns based on semantic routing for future networks.¶
In June 2021, the "Semantic Addressing and Routing for Future Networks" (SARNET-21) workshop was held as part of the IEEE International Conference on High-Performance Switching and Routing. This event was held in Paris, France and online. The main goal of the SARNET-21 workshop was to explore the use cases and network requirements in the domain of semantic addressing and routing, and identify potential research challenges to be tackled in the future. The participants and audience of the workshop were drawn chiefly from the research community.¶
This document is a report on the proceedings of the SARNET-21 workshop compiled by the authors. It captures the views and positions of the workshop participants as expressed during the workshop.¶
The workshop&s goal was to invite the research community to collectively explore semantic addressing and routing and identify potential requirements and networking solutions. Researchers and experts from industry and academia got the opportunity to share their experiences and achievements by addressing the challenges mentioned above.¶
The workshop also served as a venue to identify problems and to discover common interests that may turn into new work or into changes in the direction of ongoing work at the IETF and/or the Internet Research Task Force (IRTF).¶
The SARNET-21 workshop received 19 submissions and accepted 10 papers based on a minimum of three peer reviews. The accepted papers were presented in 3 technical sessions. The accepted papers are listed in Appendix B. In addition to the technical sessions, a keynote talk on "The Routing Challenges for Future Networks" was given by Prof. Olivier Bonaventure from Universite Catholique de Louvain (UCLouvain), and a panel discussion on "Semantic Addressing and Routing Impact on Future Networks" was moderated by Prof. Alex Galis from University College London (UCL).¶
The papers that were presented in the technical sessions cover the following topics:¶
A panel of experts was convened and chaired by Prof. Alex Galis:¶
The panelists and audience had a fascinating and constructive discussion for about 2 hours. The panelists, in their presentations, raised essential points including:¶
During the session, the panelists expressed the opinion that although semantic routing also exists today to a certain degree (e.g., ECMP, IP Flex-algo, etc.), any change that will question decades of IP network operation will undoubtedly make operators nervous. Any new addressing/routing system and framework will need to coexist with legacy gear and thus requires a robust standardization effort with one of the focal points on avoiding semantic leakage between routing domains/limited domains/slices/partitions on the Internet.¶
While this is a fundamental topic, the panel recognized that the potential impacts and benefits of changing the addressing and routing system have been overlooked. There is, hence, a need to revisit the origins of the Internet. Given that the Internet architecture is composed of many limited domains interconnected by a transport layer and associated protocols, the panelists said that understanding what is happening with limited domains and why they exist will help us better understand the impact of semantic addressing and routing. New protocols could be deployed and evaluated starting from such limited domains or network slices, where the effect is contained. Due to its fragility, there was general consensus that care needs to be taken over any changes in the backbone of the Internet.¶
As semantics is mainly service-specific standardized dynamic links between forwarding plane and services are instrumental. As such standardized, safe deterministic network programmability and frameworks enabling semantic addressing and new routing protocols with guarantees are needed, also enabling cost-efficient solutions of (In-time and On-time) service-inferred performance management.¶
Another conclusion of the panel was there is a need to explore new routing technologies and protocols inside limited domains to deliver new capabilities and better QoS, enabling application-level innovation and precluding pollution between domains. There was an observation that we are already re-using many protocols in limited domains that are currently used in the Internet, e.g., TCP/IP, BGP, etc. While this is not a bad practice, it is necessary to clearly state their "private" nature. For instance, BGP was used for routing in SIP, but it was given a new protocol ID. The panelists said that if we design protocol extensions or entirely new protocols, we need to make them compatible to avoid distorting the Internet routing system.¶
In terms of standardization, the panel concluded that researchers should be encouraged to communicate and exchange their research results. Doing research in silos doesn't help the adoption of new technologies and protocols. Fragmenting the problem and farming out the necessary engineering work to some existing silos standardization groups doesn't solve such systemic challenges either. While permissionless innovation is a way to achieve agile development of new ideas and see them deployed in limited domains, the panel's view was that mutual review is essential for stable protocols, and widescale adoption is dependent on standardization. The opinion that the IRTF is an effective and suitable place to facilitate this kind of discussion and progress was expressed.¶
Semantic addressing and routing is an exciting topic attracting a lot of attention from the research community. New semantics have been invented and deployed in limited domains for new capabilities, better QoS, higher flexibility, and efficiency. This enables incremental deployment of new technologies on "isolated islands" for innovative solutions that may or may not percolate to the whole Internet at a later stage. However, it is challenging to securely and seamlessly connect a limited domain that uses new semantic addressing and routing capabilities to the Internet. How will the new semantics will be treated in the Internet? What if different devices have different semantic routing schemes? Although non-scalable patch solutions could be used to solve this issue to some extent, this is more of a research problem rather than an engineering issue. A "holistic" approach would be to look for potential architectural patterns or common building blocks to facilitate the interconnection between the limited domain and Internet.¶
To conclude, the area of semantic addressing and routing deserves further research. The output of this research can be published and presented at future workshops, but equally important is the standardization effort that should be invested to ensure stability, scalability, and interoperability of potential solutions.¶
Based on the open discussions among the workshop participants and our overall experience with the workshop, several observations have been made, which could lead to some actions as follows:¶
This document is a workshop report and has no direct impact on security. Many of the ideas in the papers and presentations discussed at the workshop could have different security impacts. Each workshop paper should be read for its own security considerations. The security consequences of semantic addressing and routing demand further research.¶
This document has no IANA actions.¶
The authors would like to thank the distinguished keynote speaker, panelists, paper presenters, workshop participants for the exciting presentations and discussions, and the Technical Program Committee members who contributed their time to provide high-quality reviews. The workshop organizers would also like to thank the HPSR-21 conference organizers for hosting the workshop in excellent facilities in Paris, France.¶
TBD¶
The following 10 papers were presented at the workshop (listed in presentation order):¶
The keynote and panel slides can be found in the following link, <https://github.com/danielkinguk/sarah/tree/main/conferences/sarnet-21>¶