IPv6 Reference Guide
Address Format
An IPv6 address is 128 bits, written as 8 groups of 4 hex digits separated by colons:
2001:0db8:0000:0001:0000:0000:0000:0001
Two shorthand rules apply: leading zeros in a group may be omitted, and one consecutive run of all-zero groups may be replaced with :::
2001:db8:0:1::1
The prefix length (like /64) works exactly like IPv4 CIDR notation — it specifies how many leftmost bits identify the network.
Key Differences from IPv4
No broadcast. IPv6 uses multicast for what IPv4 did with broadcast. ff02::1 = all nodes on link.
/64 is the standard subnet size. The low 64 bits are the interface identifier. Subnets smaller than /64 break SLAAC autoconfiguration.
SLAAC (Stateless Address AutoConfiguration) lets hosts generate their own address from a router-advertised prefix — no DHCP required.
Multiple addresses per interface are normal: a link-local, a global or ULA unicast, and often a temporary privacy address, all at once.
No NAT needed — the address space is large enough that every device can have a globally unique address.
Well-Known Address Ranges
| Range | Name | Description |
|---|---|---|
| ::1/128 | Loopback | Equivalent to IPv4 127.0.0.1. Refers only to the local host. RFC 4291 |
| ::/128 | Unspecified | Used as source address before a real address is assigned (during DAD / DHCP). Never used as a destination. RFC 4291 |
| fe80::/10 | Link-Local Unicast | Automatically assigned to every IPv6-capable interface the moment it comes up. Not routable beyond the local segment. Used for Neighbor Discovery (NDP), SLAAC, and router/host communication on the link. Scope: link. RFC 4291 |
| fc00::/7 (fc00::/8 & fd00::/8) |
Unique Local (ULA) | The IPv6 equivalent of RFC 1918 private space (10.x, 172.16–31.x, 192.168.x). fd00::/8 is for locally-generated prefixes: pick a random 40-bit "Global ID" to form your /48 (e.g. fd12:3456:789a::/48). Scope: organization — not routable on the public internet but routable within your own network. RFC 4193 |
| ff00::/8 | Multicast | All addresses starting with ff. Common groups: ff02::1 all nodes on link, ff02::2 all routers on link, ff02::fb mDNS, ff02::1:2 DHCPv6 servers/relays. The 4-bit scope field (nibble after ff0) controls how far the packet is forwarded. RFC 4291 |
| 2000::/3 | Global Unicast (GUA) | Globally routable addresses — equivalent to public IPv4 addresses. All current ISP and RIR allocations come from this block. Top 3 bits are 001. If you have a public IPv6 address, it's almost certainly here (e.g. 2600::, 2a00::, 2001::). RFC 4291 |
| 2001:db8::/32 | Documentation | Reserved only for examples, RFCs, documentation, and training materials. Must never appear in live traffic. (Used throughout this tool's examples.) RFC 3849 |
| 2001::/32 | Teredo | Legacy IPv6-over-IPv4 UDP tunneling mechanism. Rarely seen in modern networks. RFC 4380 |
| 2002::/16 | 6to4 | Automatic IPv6-over-IPv4 tunneling. Largely deprecated; avoid in new deployments. RFC 3056 |
| ::ffff:0:0/96 | IPv4-Mapped | Represents an IPv4 address in an IPv6 socket. Used internally by dual-stack OS APIs; not normally seen on the wire. Example: ::ffff:192.0.2.1. RFC 4291 |
| 64:ff9b::/96 | NAT64 | Used by NAT64 gateways in IPv6-only networks to represent IPv4 destinations. The last 32 bits carry the IPv4 address. RFC 6052 |
| 100::/64 | Discard (IETF) | Traffic to this prefix is intentionally discarded by core routers. Useful for blackhole routing in testing. RFC 6666 |
IPv6 on Your Local Network
For a private LAN, home lab, or enterprise internal network, you have two practical addressing options:
Option 1 — Unique Local (ULA): Generate a random fd-prefix /48. Divide into /64 subnets, one per VLAN or segment. These are stable, globally unique within your org, and will never conflict with public addresses. Recommended for most private deployments where you don't have ISP-delegated IPv6 yet.
Option 2 — ISP-delegated Global Unicast: Most ISPs offer a /48 or /56 via DHCPv6-PD (Prefix Delegation). Subnet it into /64s. This gives real public addresses to every device — no NAT needed.
Regardless of option: always use /64 for end-user segments. This is the only prefix length that supports SLAAC and is expected by all modern operating systems.
You'll still see link-local addresses (fe80::) on every interface automatically — these are normal and required for the network to function, even if you don't use them for data traffic.
Common Prefix Lengths
| Prefix | Typical Meaning / Use |
|---|---|
| /32 | ISP or LIR allocation from a Regional Internet Registry (ARIN, RIPE, etc.). Contains 65,536 possible /48 sites. |
| /48 | Standard per-site or per-organization prefix. Contains 65,536 /64 subnets — plenty for any enterprise. RFC 6177 |
| /56 | Common residential ISP delegation (Comcast, etc.). Contains 256 /64 subnets — enough for most home setups. |
| /64 | The standard single subnet size. Required for SLAAC. Contains 264 ≈ 18.4 quintillion addresses. Use this for every LAN segment. |
| /127 | Point-to-point links between routers (2 addresses). Preferred over /126 in modern networks. RFC 6164 |
| /128 | A single host address — used for loopback interfaces, anycast assignments, or host-specific routes in routing tables. |
Anatomy of a Global Unicast /64 Address
When an ISP delegates a /48 to your organization and you subnet it to /64, each address has this structure:
bits 0–47
bits 48–63 (16 bits = 65,536 subnets)
bits 64–127 (auto or manual)
The interface identifier (low 64 bits) is generated by the host, either via EUI-64 (derived from the MAC address), SLAAC privacy extensions (random, rotated periodically — RFC 4941), or manual/DHCPv6 assignment.
Windows, macOS, and Linux all default to randomized interface IDs to prevent tracking devices across networks and time. This is why you'll see temporary addresses alongside the more stable ones.
ULA addresses follow the same /48 prefix : /16 subnet : /64 interface ID structure — the only difference is the fd prefix instead of a globally-routed one.