IPv6 is version 6 of the Internet Protocol, the network-layer protocol that identifies interfaces and routes packets across the Internet using 128-bit addresses. In Internet addressing it is the successor to IPv4, designed to end address exhaustion, restore end-to-end reachability and simplify packet handling in routers.
The protocol is specified mainly in RFC 8200, with addressing defined in RFC 4291. It is not backward compatible with IPv4 on the wire, so the two protocols run side by side in a dual-stack Internet, and translation or tunneling mechanisms connect hosts that speak only one of them.
How IPv6 addresses are structured
An IPv6 address is a 128-bit identifier written as eight groups of four hexadecimal digits separated by colons, such as 2001:db8:0:0:0:0:0:1. Leading zeros in a group may be dropped, and one run of zero groups may be replaced by a double colon, giving 2001:db8::1.
A typical global unicast address splits into a routing prefix, a subnet ID and a 64-bit interface identifier. A provider usually delegates a /48 or /56 to a site, and each LAN segment receives a /64 subnet, which is the standard size that stateless autoconfiguration expects.
| Address type | Prefix | Scope and purpose |
|---|---|---|
| Global unicast | 2000::/3 | Publicly routable addresses on the Internet |
| Unique local | fc00::/7 | Private site addressing, not routed globally |
| Link-local | fe80::/10 | Valid on one link only, required on every interface |
| Multicast | ff00::/8 | One-to-many delivery, replaces broadcast |
| Loopback | ::1/128 | The local host itself |
| Documentation | 2001:db8::/32 | Reserved for examples and manuals |
How IPv6 differs from IPv4
The most visible change is the address space: about 3.4 × 10^38 addresses instead of roughly 4.3 billion. The less visible changes matter just as much, because they remove work from routers and push it to the endpoints.
| Property | IPv4 | IPv6 |
|---|---|---|
| Address length | 32 bits | 128 bits |
| Base header | 20-60 bytes, variable | 40 bytes, fixed |
| Header checksum | Present | Removed, left to link and transport layers |
| Fragmentation | Routers and hosts | Source host only |
| Address resolution | ARP broadcast | Neighbor Discovery over ICMPv6 multicast |
| Broadcast | Supported | Not supported, multicast instead |
| Minimum link MTU | 68 bytes | 1,280 bytes |
Optional features move into extension headers chained through the Next Header field, such as Hop-by-Hop Options, Routing, Fragment and the IPsec headers. Routers forward the fixed base header quickly, but some networks drop packets with unusual extension headers, which is a known deployment friction.
Because routers never fragment, IPv6 relies on Path MTU Discovery through ICMPv6 Packet Too Big messages. A firewall that blocks all ICMPv6 therefore breaks large transfers, a failure that many IPv4-trained administrators do not expect.
Address assignment and neighbor discovery
IPv6 hosts configure themselves through Neighbor Discovery Protocol (NDP), which uses ICMPv6 messages for router discovery, address resolution and duplicate address detection. Routers send Router Advertisements that announce the prefix, the default gateway and flags telling hosts which assignment method to use.
| Method | How the host gets an address | Typical use |
|---|---|---|
| SLAAC | Combines the advertised /64 prefix with a self-generated interface ID | Home networks, mobile devices |
| Stateful DHCPv6 | Server leases specific addresses and options | Enterprises needing address records |
| Stateless DHCPv6 | SLAAC address plus DNS and other options from a server | Mixed environments |
| DHCPv6 prefix delegation | Router receives a whole prefix to subdivide | ISP to customer router |
| Static | Configured manually | Servers, routers, infrastructure |
Early SLAAC derived the interface ID from the MAC address (EUI-64), which exposed a stable hardware identifier across networks. Modern systems use temporary privacy addresses and stable opaque identifiers instead, so tracking a device by its address is much harder.
Transition and coexistence with IPv4
No flag day replaced IPv4, so networks use coexistence mechanisms chosen by which protocol the access network and the destination support. The preferred long-term model is IPv6-only access with IPv4 delivered as a service.
| Mechanism | Principle | Where it fits |
|---|---|---|
| Dual stack | Hosts and routers run both protocols natively | Most enterprise and ISP networks today |
| NAT64 with DNS64 | Translates IPv6-only clients to IPv4 servers | Mobile carriers, IPv6-only data centers |
| 464XLAT | Adds client-side translation for IPv4-only apps | Mobile networks on top of NAT64 |
| DS-Lite | Tunnels IPv4 inside IPv6 to a carrier NAT | Cable and fiber providers |
| 6in4 tunnels | Encapsulates IPv6 inside IPv4 | Sites without native IPv6 transit |
Clients that see both A and AAAA records use Happy Eyeballs logic, trying IPv6 first and falling back to IPv4 after a short delay of about 50-300 milliseconds. This hides broken IPv6 paths from users but can also mask misconfiguration for months.
Common mistakes when deploying IPv6
- Assuming NAT equals security. IPv6 normally has no NAT, so a stateful firewall must enforce the same inbound policy that NAT used to imply.
- Blocking all ICMPv6. Neighbor Discovery and Packet Too Big messages are essential; filter selectively instead.
- Using prefixes longer than /64 on LANs. SLAAC and many devices expect exactly a /64.
- Forgetting rogue Router Advertisements. Enable RA Guard on access switches to stop unauthorized gateways.
- Leaving IPv6 enabled but unmanaged. Operating systems prefer IPv6 by default, so an unmonitored stack becomes an unseen attack path.
The idea in brief
Every device on the Internet needs an address so that data can find it, much like every house needs a postal address. The older system, IPv4, has run out of addresses, so networks share one address among many devices.
IPv6 is the newer addressing system with so many addresses that every device can have its own. It also lets devices set up their addresses automatically. The key point is simple: IPv6 is the long-term address system of the Internet, and it works alongside IPv4 until the old one is no longer needed.
A practical scenario
A regional Internet provider receives a /32 IPv6 allocation and gives each fiber customer a /56 prefix through DHCPv6 prefix delegation. The customer router splits it into separate /64 subnets for the home network, a guest network and smart-home devices, while hosts configure themselves with SLAAC and privacy addresses.
IPv4 is still delivered through carrier-grade NAT, but popular streaming, search and social platforms answer over native IPv6. As a result, most traffic bypasses the provider’s NAT devices, which reduces logging load and delays costly NAT capacity upgrades, while customers keep a stateful firewall on the router that blocks unsolicited inbound connections.