IPv6 deployment
/ eye-pee-vee-six /
Every device on the Internet needs an address, like a postal address for packets. The old scheme, IPv4, has about 4.3 billion addresses (2^32) — which sounded endless in the 1980s but ran out as billions of phones, laptops, and gadgets came online. IPv6, the successor, has a practically inexhaustible supply (2^128 addresses — enough to give every grain of sand on Earth its own range many times over). IPv6 deployment is the decades-long, still-unfinished project of actually switching the world over to this new addressing — and it is a vivid lesson in how hard it is to upgrade something as enormous and decentralized as the Internet.
Concretely, IPv6 is not a tweak to IPv4; it is a separate, incompatible addressing scheme, so a pure-IPv6 host and a pure-IPv4 host cannot talk directly. There is no flag day where everyone switches at midnight — no one is in charge of the whole Internet. Instead the migration runs through a long transition period in which both protocols coexist: hosts run dual-stack (speaking both IPv4 and IPv6 at once), and various translation and tunneling mechanisms bridge the two worlds. Each network operator, content provider, and device vendor must upgrade on their own schedule, and a service only gains the benefit of IPv6 once both ends and the path between them support it.
Why it matters and the honest reality: IPv6 has been standardized since the 1990s, yet adoption took decades and is still only partial — a striking example of inertia in a system with no central authority and huge installed bases. The main stopgap that took the pressure off was network address translation (NAT), which lets many devices share one public IPv4 address; NAT bought time but also broke the clean end-to-end model and added complexity, which ironically reduced the urgency to move to IPv6. A common misconception is that IPv6 is mainly about more addresses; it also cleans up the header and removes the need for NAT — but its slow, grinding rollout is the real story, and a reminder that 'the right design' does not deploy itself.
A modern phone typically runs dual-stack: when it opens a website, it tries the site's IPv6 address and its IPv4 address and uses whichever connects first. If the website, the phone's network, and every hop between them all support IPv6, the connection rides IPv6; if any link in the chain is IPv4-only, it quietly falls back to IPv4 — often through a carrier's NAT.
Dual-stack lets old and new addressing coexist during the long transition.
IPv6 deployment shows that a better design does not deploy itself: with no central authority and billions of installed devices, migration crawls. NAT, the IPv4 stopgap, worked so well it sapped the urgency — which is why decades on, the Internet still runs both protocols side by side.