IPv6 addressing: why it exists, notation and shortening, address types, prefixes, SLAAC and dual stack

IPv4 has about 4.3 billion addresses, and they've run out. IPv6 (Internet Protocol version 6) fixes this with 128-bit addresses: about 340 undecillion (3.4 × 10³⁸) of them, enough to give every device on Earth trillions of addresses. Major mobile networks, Google, Facebook, Netflix and cloud providers already run IPv6, and in many countries a large share of users reach Google over IPv6. Kenyan mobile and fibre operators have been rolling it out too. Network engineers must be comfortable reading, writing and planning IPv6.

Why IPv6?

IPv4 and IPv6 aren't directly compatible, so the world runs both side by side (dual stack) during a long transition.

Writing IPv6 addresses

128 bits are written as 8 groups of 4 hexadecimal digits (16 bits each), separated by colons:

2001:0db8:0000:0000:0000:ff00:0042:8329

(The 2001:db8::/32 range is reserved for documentation and examples.)

The two shortening rules

Rule 1: drop leading zeros in each group.

2001:0db8:0000:0000:0000:ff00:0042:8329
2001:db8:0:0:0:ff00:42:8329

**Rule 2: replace ONE run of consecutive all-zero groups with ::.**

2001:db8:0:0:0:ff00:42:8329
2001:db8::ff00:42:8329

Important details:

• :: can appear only once in an address; otherwise you couldn't tell how many zeros each one hides.
• Only leading zeros are dropped: 0db8 → db8, but db80 stays db80.
• If there are two equal runs of zeros, shorten the first (the standard, RFC 5952, says shorten the longest, or the first if tied), and use lowercase letters.

To expand an address, count the groups present and fill :: with enough 0000 groups to make 8.

import ipaddress
for a in ["2001:0db8:0000:0000:0000:ff00:0042:8329", "2001:db8:0:1::5", "fe80::21a:2bff:fe3c:4d5e", "::1"]:
    ip = ipaddress.ip_address(a)
    print(f"{a:42} short: {ip.compressed:28} full: {ip.exploded}")

Prefixes

IPv6 uses CIDR prefixes like IPv4. An address is usually split in half:

2001:db8:abcd:0012 : 0000:0000:0000:0001
|-- network prefix (64) --||-- interface ID (64) --|

Subnetting IPv6 is usually easier than IPv4: you just change the 4th group. From 2001:db8:abcd::/48, subnets are 2001:db8:abcd:0001::/64, 2001:db8:abcd:0002::/64, ... up to ffff. You never need to calculate "usable hosts": a /64 has 2⁶⁴ addresses (18 quintillion).

Address types

Useful multicast groups: ff02::1 (all nodes on the link), ff02::2 (all routers on the link).

There's no broadcast in IPv6. Tasks that used broadcasts (like ARP) now use targeted multicast, so devices aren't disturbed by traffic that isn't for them.

How devices get IPv6 addresses

With SLAAC, the interface ID used to be built from the MAC address (EUI-64: insert fffe in the middle and flip one bit). Modern operating systems use random, temporary interface IDs for privacy, so websites can't track a device by its address.

A device typically has several IPv6 addresses at once: a link-local (fe80::...), one or more global addresses, and temporary privacy addresses. That's normal.

Neighbour Discovery Protocol (NDP)

NDP (part of ICMPv6) replaces ARP and more:

• Neighbour Solicitation/Advertisement: find a neighbour's MAC address (like ARP).
• Router Solicitation/Advertisement: find routers and prefixes (for SLAAC).
• Duplicate Address Detection: check nobody else uses an address before taking it.

Because NDP uses ICMPv6, blocking all ICMPv6 on a firewall breaks IPv6. Filter it carefully instead.

Transition: dual stack, tunnels and translation

When both work, devices prefer IPv6 (the "Happy Eyeballs" method races both and uses the faster).

Seeing and testing IPv6

ipconfig                 # Windows: look for "IPv6 Address" and "Link-local IPv6 Address"
ip -6 addr               # Linux
ping -6 google.com       # Windows/Linux: test IPv6 connectivity
ping ::1                 # loopback test
tracert -6 google.com    # Windows path over IPv6 (Linux: traceroute -6)
nslookup -type=AAAA google.com   # IPv6 address records

Many "test my IPv6" websites show whether your connection supports it.

Writing an IPv6 plan

Example: an ISP gives a school 2001:db8:5a00::/48.

• Lab 1: 2001:db8:5a00:10::/64
• Lab 2: 2001:db8:5a00:11::/64
• Staff: 2001:db8:5a00:20::/64
• CCTV: 2001:db8:5a00:40::/64
• Wi-Fi: 2001:db8:5a00:30::/64

Matching the subnet number to the IPv4 VLAN/third octet makes dual-stack networks easy to read.

Summary

• IPv6 uses 128-bit addresses written as 8 groups of hex; it removes the need for NAT and broadcasts.
• Shorten by dropping leading zeros and replacing one run of zero groups with :: (only once).
• LANs use /64; sites get /48 or /56; subnet by changing the 4th group.
• Types: global unicast (2000::/3), link-local (fe80::/10), unique local (fd00::/8), loopback ::1, multicast ff00::/8.
• Addresses come from SLAAC, DHCPv6 or static configuration; NDP replaces ARP; networks run dual stack.

Learn more: Google IPv6 statistics · Cloudflare: What is IPv6?