Filtrer par genre

PING is a podcast for people who want to look behind the scenes into the workings of the Internet. Each fortnight we will chat with people who have built and are improving the health of the Internet. The views expressed by the featured speakers are their own and do not necessarily reflect the views of APNIC.

77 - The IPv6 Transition
0:00 / 0:00
1x
  • 77 - The IPv6 Transition

    In his regular monthly spot on PING, APNIC’s Chief Scientist Geoff Huston (https://blog.apnic.net/author/geoff-huston/) discusses the slowdown in worldwide IPv6 uptake. Although within the Asia-Pacific footprint we have some truly remarkable national statistics, such as India which is now over 80% IPv6 enabled (https://stats.labs.apnic.net/ipv6/IN)by APNIC Labs measurements, And Vietnam which is not far behind on 70% (https://stats.labs.apnic.net/ipv6/VN) the problem is that worldwide, adjusted for population and considering levels of internet penetration in the developed economies, the pace of uptake overall has not improved and has been essentially linear since 2016 (https://stats.labs.apnic.net/ipv6/XA). In some economies like the US, a natural peak of around 50% capability was reached in 2017 (https://stats.labs.apnic.net/ipv6/US) and since then uptake has been essentially flat: There is no sign of closure to a global deployment in the US, and many other economies.Geoff takes a high level view of the logisitic supply curve with the early adopters, early and late majority, and laggards, and sees no clear signal that there is a visible endpoint, where a transition to IPv6 will be "done". Instead we're facing a continual dual-stack operation of both IPv4 (increasingly behind Carrier Grade Nats (CGN) deployed inside the ISP) and IPv6.There are success stories in mobile (such as seen in India) and in broadband with central management of the customer router. But, it seems that with the shift in the criticality of routing and numbering to a more name-based steering mechanism and the continued rise of content distribution networks, the pace of IPv6 uptake worldwide has not followed the pattern we had planned for.Read more about the IPv6 transition at the APNIC Blog* The IPv6 Transition (https://blog.apnic.net/2024/10/22/the-ipv6-transition/) (Geoff Huston, APNIC Blog November 2024)* The Transition to IPv6 are we there yet (https://blog.apnic.net/2022/05/04/the-transition-to-ipv6-are-we-there-yet/) (Geoff Huston, APNIC Blog May 2022)

    Wed, 13 Nov 2024 - 59min
  • 76 - A student-led IPv6 deployment at NITK Karnataka

    In this episode of PING, Vanessa Fernandez and Kavya Bhat, two students from the National Institute of Technology Karnataka (NITK) (https://www.nitk.ac.in/) discuss the student led, multi-year project to deploy IPv6 at their campus. Kavya & Vanessa have just graduated, and are moving into their next stages of work and study in computer sciences and network engineering.Across 2023 and 2024 they were able to attend IETF118 and IETF119 and present on their project and it’s experiences to the IPv6 working groups and off-Working Group meetings, in part funded by the APNIC ISIF Project and the APNIC Foundation.This multi-year project is supervised by the NITK Centre for Open-source Software and Hardware (COSH) and has outside review from Dhruv Dhody (ISOC) and Nalini Elkins (Inside Products inc). Former students have also acted as alumni and remain involved in the project as it progresses.We often focus on IPv6 deployment at scale in the telco sector, or experiences with small deployments in labs, but another side of the IPv6 experience is the large campus network, in scale equivalent to a significant factory or government department deployment but in this case undertaken by volunteer staff, with little or no prior experience of networking technology. Vanessa and Kavya talk about their time on the project, and what they got to present at IETF.Read more information on the NITK and their IPv6 deployment project on the APNIC Blog, the IETF website and the APNIC Foundation pages:* Migrating the NITK Surathkal Campus Network to IPv6 (https://apnic.foundation/projects/migrating-nitk-surathkal-campus-network-to-ipv6/) (APNIC Foundation)* How Deploying IPv6 at NITK Led me to IETF (https://blog.apnic.net/2024/07/08/how-deploying-ipv6-at-nitk-led-me-to-ietf/) (Vanessa Fernandez, APNIC Blog)* IPv6 Deployment at NITK (https://datatracker.ietf.org/meeting/118/materials/slides-118-v6ops-ipv6-deployment-at-nitk-00) (IETF118 Presentation)

    Wed, 30 Oct 2024 - 27min
  • 75 - The back of the class: looking at 240/4 reachability

    In his regular monthly spot on PING, APNIC’s Chief Scientist, Geoff Huston (https://blog.apnic.net/author/geoff-huston/), discusses a large pool of IPv4 addresses left in the IANA registry, from the classful allocation days back in the mid 1980s. This block, from 240.0.0.0 to 255.255.255.255 encompasses 268 million hosts, which is a significant chunk of address space: it's equivalent to 16 class-A blocks, each of 16 million hosts. Seems a shame to waste it, how about we get this back into use?Back in 2007 Geoff Paul and myself submitted An IETF Draft (https://datatracker.ietf.org/doc/draft-wilson-class-e/history/) which would have removed these addresses from the "reserved" status in IANA and used to supplement the RFC1918 private use block. We felt at the time this was the best use of these addresses because of their apparent un-routability, in the global internet. Almost all IP network stacks at that time shared a lineage with the BSD network code developed at the University of California, and released in 1983 as BSD4.2. Subsequent versions of this codebase included a 2 or 3 line rule inside the Kernel which checked the top 4 bits of the 32 bit address field, and refused to forward packets which had these 4 bits set. This reflected the IANA status marking this range as reserved. The draft did not achieve consensus.A more recent proposal has emerged (https://datatracker.ietf.org/doc/draft-schoen-intarea-unicast-240/) from Seth Schoen, David Täht and John Gilmore in 2021 which continues to be worked on, but rather than assigning to RFC1918 internal non-routable puts the address into global unicast use. The authors believe that the critical filter in devices has now been lifted, and no longer persists at large in the BSD and Linux derived codebases. This echoes use of the address space which has been noted inside the Datacentre.Geoff has been measuring reachability at large to this address space, using the APNIC Labs measurement system and a prefix in 240.0.0.0/4 temporarily assigned and routed in BGP. The results were not encouraging, and Geoff thinks routability of the range remains a very high burden.Read more about 240/4 in the APNIC Blog, and the IETF Datatracker website:* Looking for 240/4 addresses (https://blog.apnic.net/2024/09/10/looking-for-240-4-addresses/) (Geoff Huston, APNIC Blog September 2024)* Re-delegation of 240/4 from "future use" to "private use" (https://datatracker.ietf.org/doc/draft-wilson-class-e/) (expired IETF draft, 2008)* Unicast use of the formerly reserved 240/4 (https://datatracker.ietf.org/doc/draft-schoen-intarea-unicast-240/00/) (active IETF draft, 2024)

    Wed, 16 Oct 2024 - 1h 09min
  • 74 - Focusing purely on technology limits the understanding of Internet resilience

    In this episode of PING, Nowmay Opalinski (https://blog.apnic.net/author/nowmay-opalinski/) from the French Institute of Geopolitics at Paris 8 University discusses his work on resilience, or rather the lack of it, confronting the Internet in Pakistan.As discussed in his blog post (https://blog.apnic.net/2024/09/17/focusing-on-technology-limits-understanding-of-internet-resilience-pakistan-case-study/), Nowmay and his colleagues at the French Institute of Geopolitics (IFG), University Paris 8 (https://www.univ-paris8.fr/en/ur-centre-de-recherches-et-d-analyses-geopolitiques-ifg-lab-research-unit), and LUMS University Pakistan (https://lums.edu.pk/) used a combination of technical measurement from sources such as RIPE Atlas (https://atlas.ripe.net/), in a methodology devised by the GEODE project, combined with interviews in Pakistan, to explore the reasons behind Pakistan’s comparative fragility in the face of seaborne fibre optical cable connectivity. The approach deliberately combines technical and social-science approaches to exploring the problem space, with quantitative data and qualitative interviews.Located at the head of the Arabian Sea, but with only two points of connectivity into the global Internet, Pakistan has suffered over 22 ‘cuts’ to the service in the last 20 years, However, as Nowmay explores in this episode, there actually are viable fibre connections to India close to Lahore, which are constrained by politics.Nowmay is completing a PhD at the institute, and is a member of the GEODE project (https://geode.science/en/about/). His paper on this study was presented at the 2024 AINTEC conference (https://interlab.ait.ac.th/aintec2024) held in Sydney, as part of ACM SIGCOMM 2024 (https://conferences.sigcomm.org/sigcomm/2024/).Read more about GEODE, and Nowmay’s work:* The GEODE project (https://geode.science/en/)* Pakistan, a case study (https://blog.apnic.net/2024/09/17/focusing-on-technology-limits-understanding-of-internet-resilience-pakistan-case-study/) in Internet fragility* The Quest for a Resilient Internet Access in a Constrained Geopolitical Environment  (https://dl.acm.org/doi/fullHtml/10.1145/3674213.3674220)(AINTEC 2024 Paper)

    Wed, 02 Oct 2024 - 34min
  • 73 - Privacy and DNS Client Subnet

    In his regular monthly spot on PING, APNIC’s Chief Scientist, Geoff Huston (https://blog.apnic.net/author/geoff-huston/), discusses another use of DNS Extensions: The EDNS0 Client Subnet option (RFC 7871 (https://datatracker.ietf.org/doc/html/rfc7871)). This feature, though flagged in its RFC as a security concern, can help route traffic based on the source of a DNS query. Without it, relying only on the IP address of the DNS resolver can lead to incorrect geolocation, especially when the resolver is outside your own ISP’s network.The EDNS Client Subnet (ECS) signal can help by encoding the client’s address through the resolver, improving accuracy in traffic routing. However, this comes at the cost of privacy, raising significant security concerns. This creates tension between two conflicting goals: Improving routing efficiency and protecting user privacy.Through the APNIC Labs measurement system (https://labs.apnic.net/measurements/), Geoff can monitor the prevalence of ECS usage in the wild. He also gains insights into how much end-users rely on their ISP’s DNS resolvers versus opting for public DNS resolver systems that are openly available.Read more about EDNS0 and UDP on the APNIC Blog and at APNIC Labs:* Privacy and DNS Client Subnet (Geoff Huston, APNIC Blog July 2024) (https://blog.apnic.net/2024/07/23/privacy-and-dns-client-subnet/)* The use of ECS as measured by APNIC Labs (https://stats.labs.apnic.net/ecs)

    Wed, 18 Sep 2024 - 49min
Afficher plus d'épisodes