IPv6 calculator

IPv6 addresses are 128 bits written as eight groups of four hex digits, and the two shortening rules are what make them look inconsistent. Leading zeros in a group may be dropped, and one run of all-zero groups may be replaced with ::. Only one run, because two would be ambiguous — there would be no way to tell how many groups each :: stood for.

RFC 5952 turned the "may" into a "must" for anything that outputs an address, and the rules are specific: lowercase hex, leading zeros always dropped, :: used on the longest run of zero groups, the leftmost when two runs are equally long, and never on a run of only one group. That last rule catches people out — 2001:db8:0:1:1:1:1:1 keeps its single zero written out.

The scope prefixes are worth recognising. fe80::/10 is link-local and every interface has one whether you configured it or not. fc00::/7 is unique local, the rough equivalent of RFC 1918. 2000::/3 is global unicast — the actually-routable space. And ff00::/8 is multicast; IPv6 has no broadcast at all, which is one of the real design changes from IPv4.

Subnetting is different in practice too. A /64 is the standard network size because stateless autoconfiguration needs 64 host bits, so the host part is not something you economise on — an ISP hands out a /56 or /48 and you cut /64s from it.