[menog] [Paper] New IP

Fahd Batayneh fahd.batayneh at icann.org
Sun Nov 29 16:23:43 UTC 2020

Friends and Colleagues,

ICANN’s Office the Chief Technology Officer (OCTO) has published a new paper entitled “New IP”; which, you can access online here >> https://www.icann.org/octo-017-en.pdf.

Network 2030 was a focus group (FG) created by the Telecommunication Standardization Sector (ITU-T) Study Group 13 “to carry out a broad analysis for future networks towards 2030 and beyond. In order to formulate a right vision, this FG is expected to identify the gaps and challenges based on the latest networking technologies, and derive fundamental requirements from novel use cases.” The Network 2030 Focus Group concluded in July 2020, envisioning a number of futuristic use cases, ranging from “holographic communications” to “tactile Internet,” “Digital Twins,” and “Industrial IoT.” The requirements perceived for these use cases demand bandwidth on the order of one terabit per second per-flow, sub-millisecond latency, and zero packet loss. These requirements seem unlikely to be ubiquitously realizable in the assumed timeframe of ten years from now.

New IP is driven by Huawei and its subsidiary, Futurewei. New IP’s relationship to Network 2030 is unclear because New IP proponents tend to use the two names interchangeably. At best, New IP can be seen as a set of desired features to implement the use case described in Network 2030. However, there are no publicly available, definitive, and complete descriptions of what New IP is. As such, it can only be seen at best as “work in progress” and cannot be fully analyzed and compared to a standard such as the TCP/IP protocol suite. Hints can be found in Huawei blogs, a Futurewei Internet Draft submitted to the Internet Engineering Task Force (IETF), slides from a guest talk at an Institute of Electrical and Electronics Engineers (IEEE) conference, and in an ITU-T liaison statement to the IETF. At a high level, New IP architecture introduces variable length addresses; reintroduces circuit-switched-like principles in what is dubbed “better than best effort networking”; suggests an approach to enable packets to embed contracts to be enforced by intermediary network elements in a way that is reminiscent of active networks where packets contain code to be executed by routers and switches; and presents the concept of “ManyNets” where instead of a single network, the Internet would become a patchwork of networks loosely interconnected via gateways. New IP advances the idea of a strong regulatory binding between an IP address and a user. If deployed, such techniques could make pervasive monitoring much easier because it would allow any intermediary element (router, switch, and so on) to have full access to exactly which user is doing what. Similarly, content providers would have access to the identity of every user connecting to them. This could dramatically increase the oversight of published content.

Although New IP can use a new variable length addressing type, IPv4, IPv6, or any combination of the above, it cannot be compatible with the existing deployed IPv4- or IPv6-based infrastructure. As such, New IP would have to be deployed in parallel with the current Internet infrastructure, interconnecting via gateways. Any significant deployment would probably face decades-long timelines.

More papers published by OCTO can be found here >> https://www.icann.org/resources/pages/octo-publications-2019-05-24-en.

Thank you,

Fahd Batayneh
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://lists.menog.org/pipermail/menog/attachments/20201129/7449854e/attachment.html 

More information about the Menog mailing list