A short history of IPv6

Every device on the Internet needs an address, like a house needs a street number. IPv6 exists because the original numbering ran short. This is how that played out.

  1. Thirty-two bits seemed like plenty

    The Internet Protocol, version 4, was written for a research network of a few hundred computers. Its addresses are 32 bits long, which allows about 4.3 billion of them. Addresses were handed out in three block sizes, called classes A, B and C, and large organisations received huge blocks they never filled.

  2. The first warning

    The Web arrived and the network started growing fast. Engineers projected that address space, especially the mid-sized blocks, would run out within a few years. They answered in two ways: stretch what existed, and design a replacement. In 1993 classless addressing (CIDR) replaced the fixed block sizes with flexible ones, which bought time.

  3. A successor takes shape

    The IETF, the body that standardises Internet protocols, asked for proposals for an "IP Next Generation", which gave this site its name. Several competing designs were weighed, and the one that won grew addresses to 128 bits and simplified the packet header. A recommendation was published in January 1995. The new protocol became version 6 because number 5 had already been used by an experimental streaming protocol.

  4. A stopgap that worked too well

    Network address translation (NAT) let many computers share a single public address, and private address ranges were set aside for use behind it. Homes and companies were happy, and the pressure to move eased. The price was that devices behind NAT could no longer be reached directly from outside, which quietly shaped how a generation of software was written.

  5. A standard and a playground

    The first IPv6 specification came out in December 1995, and a revised one followed in 1998. In 1996 volunteers and research labs linked up the 6bone, an experimental IPv6 network built from tunnels running over IPv4. It was switched off on 6 June 2006, once real IPv6 service made it unnecessary.

  6. The first real address blocks

    The regional registries that distribute addresses began receiving IPv6 space and passing it on to providers and organisations.

  7. The last IPv4 blocks

    In February, the central pool of IPv4 addresses handed out its final blocks. The regional registries ran out one after another over the following years. In June, major websites joined World IPv6 Day and ran IPv6 for 24 hours as a public test.

  8. Switched on for good

    On World IPv6 Launch Day, network providers, home router makers and big websites left IPv6 enabled permanently. The "IPv6 enabled" badges you still see on websites date from this campaign.

  9. A full Internet Standard

    The IPv6 specification, RFC 8200, was promoted to Internet Standard, more than twenty years after the first version. You can read it on this site.

  10. Two protocols, side by side

    The share of users reaching services over IPv6 rises every year, and Google publishes the figures. Mobile operators and large content providers lead the way, while many companies still run both protocols. Meanwhile IPv4 addresses are bought, sold and rented, which shows how scarce they have become.

Why did it take so long?

No direct bridge

IPv4 and IPv6 cannot talk to each other without translation, so everything had to run both for a long time.

The workaround worked

NAT solved the immediate shortage, so the benefit of switching was long term, while the cost was immediate.

Who goes first?

Websites waited for users with IPv6, and providers waited for websites that needed it.