IPv4-with-IPv6 Next-Hop: Towards an IPv4-Free Internet Core#

Tobias Fiebig — Vienna University of Technology

• Why: For anyone who cares about BGP/datapath internals and the long, slow IPv6 transition, this is one of the sharpest technical titles on the program. The question it names — can we carry IPv4 reachability with IPv6-only transport in the core, so the core stops speaking IPv4 entirely — is one of the few genuinely structural changes still available to the internet without a flag day.
• What: The title indicates a proposal or evaluation of running the core network IPv6-only: advertising IPv4 prefixes while using IPv6 next-hop addresses (the mechanism standardised in RFC 8950), eliminating the need for IPv4 addresses on core-facing interfaces and shrinking the core's IPv4 configuration surface. No abstract was published, so the exact contribution — measurement of current adoption, an operator how-to, or a transition proposal — cannot be confirmed. (from abstract only; no description published)
• How: Presumably by combining extended-nexthop BGP encoding (IPv4 NLRI, IPv6 next hop) with v6-only core links, so IPv4 survives only as payload/edge reachability. Fiebig's group is known for large-scale BGP measurement, so expect deployment data or adoption obstacles rather than a purely theoretical argument.
• Where applicable: Transit providers, IXs, and large DC/WAN interconnect operators — exactly the environments this event's audience runs. The main constraints are real-world ones: hardware/daemon support for extended nexthops on all core routers, and the operational discipline to keep v4 out once it's out. Treat the specifics as unverified until the paper or slides are published.