What you can use it for
With IPv6 you’re rarely short of addresses. Planning is about keeping networks organised, handing out prefixes cleanly and catching notation mistakes.
Numbering VLANs inside a /48
Say your team is given 2001:db8:1200::/48 and needs LANs for staff, servers, voice, Wi-Fi, building systems and guests. The 16 bits between /48 and /64 give you 65,536 possible /64s, so you can make the subnet ID match the VLAN number. VLAN 10 becomes 2001:db8:1200:10::/64, VLAN 20 becomes 2001:db8:1200:20::/64, and so on. Put the /48 into the planner to list them and confirm none clash.
Handing out prefixes to sites or customers
Big allocations are usually split in layers. An ISP might give each customer a /56, which holds 256 LANs. A company with a /40 might give each region a /48, which holds 65,536. A branch that receives 2001:db8:4500:7a00::/56 can number its /64s from 7a00 to 7aff. The calculator shows where each block starts and ends, so you don’t give two sites overlapping space.
Troubleshooting a device’s address
Enter something like 2001:db8:abcd:12:4a2f:10ff:fe91:7c20/64 and the calculator splits it into its subnet, 2001:db8:abcd:12::/64, and the 64-bit interface ID. That tells you whether the device landed on the right LAN and whether two hosts can talk without a router. The expanded form also shows up missing or misplaced groups, and the address type and reverse-DNS name help when you’re setting up DNS or tracking down a link-local versus global address mix-up.
IPv6 prefixes vs IPv4 subnet masks
An IPv6 address is 128 bits long, and the network boundary is always written as a prefix length such as /64. There’s no dotted subnet mask and no broadcast address. Jobs that IPv4 handles with broadcast are done with multicast and neighbour discovery instead.
Why /64 is the standard LAN size
Ordinary LANs, whether global or unique-local, use a /64, which leaves 64 bits for the interface ID. That’s far more addresses than any LAN will use, and it’s meant to be. Features like SLAAC rely on it, so you size IPv6 networks by structure, not by counting devices.
Site prefixes and subnet IDs
A /48 leaves 16 bits of subnet ID before the /64 boundary, enough for 65,536 LANs. A /56 leaves 8 bits, which gives you 256.
Stick to nibble boundaries where you can
Each hex digit in an IPv6 address is four bits, known as a nibble. If your prefix length is a multiple of four, like /48, /52, /56 or /60, the boundary falls neatly between two digits. You can read the network straight off the address, and every subnet ID uses whole digits.
It matters most for reverse DNS, where each hex digit becomes one label in ip6.arpa. Say a region is handed 2001:db8:40::/46. That boundary sits halfway through a digit, so there’s no single reverse zone for it. You’d need four: one each for the /48s from 2001:db8:40:: to 2001:db8:43::. Round to a multiple of four and you avoid that juggling.
Common IPv6 prefix sizes
| Prefix | /64s inside | Typically used for |
|---|---|---|
| /32 | 4,294,967,296 | An ISP’s allocation |
| /40 | 16,777,216 | A large organisation or a provider’s region |
| /48 | 65,536 | One site or organisation |
| /56 | 256 | A home or small business connection |
| /60 | 16 | A smaller customer delegation |
| /64 | 1 | One LAN |
| /127 | — | A link between two routers |
| /128 | — | One address, such as a loopback |
A /127 or /128 is smaller than a LAN, so it holds no /64s. They’re only for router links and single addresses, never for a network of devices.
Compressed and expanded notation
You can drop leading zeros in any group, and replace one run of all-zero groups with ::. The expanded form always shows all eight groups at four hex digits each.