IPv6-only networks still carry IPv4 subnets and ARP as "1982-era baggage" because applications and legacy systems expect IPv4. This operational liability includes ARP storms, cache poisoning, table exhaustion, and address waste—a /30 link wastes half its addresses. Large providers like Hetzner, OVHcloud, and Scaleway already bypass this with proprietary /32 gateway workarounds, but their solutions are non-interoperable.

The IETF proposal draft-vanmook-intarea-ipv6-resolved-gateway, born from a RIPE 91 lightning talk in Bucharest and now in its third version with a presentation at IETF 126 in Vienna, defines a single sentinel address: 192.0.0.11. A DHCPv4 server hands it out as the standard Router Option (Option 3). Updated hosts recognise the sentinel and resolve the router's MAC from their IPv6 neighbour cache instead of ARPing. Unmodified hosts simply ARP for it, and the router answers—both coexist indefinitely.

The mechanism slots into RFC 8950 (IPv4 prefixes with IPv6 next-hops) and draft-ietf-intarea-v4-via-v6, forming a single-stack architecture serving dual-stack endpoints. IPv4 becomes a service identifier over IPv6-only transport, eliminating subnet sprawl, reducing attack surface, and collapsing provider-specific gateways into one universal address.