IPv6 Addressing and Prefixes

What it is

IPv6 is the newer version of the Internet Protocol, and its main change is a 128-bit address instead of 32 bits.

That gives an enormous number of addresses, so every device can have its own globally unique one without the workarounds IPv4 needs.

IPv6 addresses are written in hexadecimal, as eight groups of four hex digits separated by colons:

2001:0DB8:0000:0001:0000:0000:0000:0001

Each group is 16 bits. The prefix length, written after a slash, says how many leading bits identify the network.

For most LANs it is /64: the first 64 bits are the network, and the last 64 bits identify the interface.

How it works

You shorten addresses with two rules, read the prefix as the network part, and configure it on an interface.

Shortening an address

Long addresses are shortened with two rules:

1. Drop leading zeros in each group. 0DB8 becomes DB8, and 0001 becomes 1.
2. Replace one run of consecutive all-zero groups with ::. You can do this only once in an address, otherwise it would be unclear how many groups each :: stands for.

Applying both rules to the address above gives:

2001:DB8:0:1::1

To expand a short address, put back the leading zeros and replace :: with enough zero groups to reach eight groups in total.

Hex digits are not case sensitive. Addresses are usually written in lowercase in practice.

Prefixes

The prefix length works the same way as in IPv4. For 2001:db8:1:1::/64, the first 64 bits, four groups, are the network, and the rest are host bits.

A common site plan gives each organisation a larger block, such as a /48, and then splits it into many /64 subnets, one per VLAN. 2001:db8::/32 is reserved for documentation, which is why you see it in examples.

To see how much room that gives, think of a /48 site block. The 16 bits between /48 and /64 form the subnet ID, so one site can create 65,536 separate /64 subnets.

Because /64 is the standard LAN size, you do not need to calculate host counts the way you do in IPv4. You simply number the subnets, for example 2001:db8:1:1::/64, 2001:db8:1:2::/64 and so on.

Prefixes are also used in routing. A route to 2001:db8:1::/48 covers every /64 subnet inside that block, in the same way an IPv4 summary route covers smaller subnets.

There is no broadcast in IPv6. Functions that IPv4 handled with broadcast use multicast instead.

Configuring an interface

Routing for IPv6 must be switched on before the router forwards IPv6 packets:

ipv6 unicast-routing
interface GigabitEthernet0/0
 ipv6 address 2001:db8:1:1::1/64
 no shutdown

You can also let the router build the interface part of the address from its MAC address, with the EUI-64 method:

interface GigabitEthernet0/0
 ipv6 address 2001:db8:1:1::/64 eui-64

A link-local address is created automatically once IPv6 is enabled on an interface, for example by configuring a global address or with ipv6 enable. You can set your own, which is easier to read:

interface GigabitEthernet0/0
 ipv6 address fe80::1 link-local

How hosts get addresses

A host can get its address in several ways:

• Static: configured by hand.
• SLAAC (Stateless Address Autoconfiguration): the host listens for a router advertisement that contains the prefix, then builds its own interface part. No server holds a record of the address.
• DHCPv6: a server hands out the address (stateful), or only extra information such as DNS servers (stateless).

Routers also help hosts find each other with the Neighbor Discovery Protocol (NDP), which replaces ARP. NDP uses ICMPv6 messages.

Verifying

show ipv6 interface brief
show ipv6 interface GigabitEthernet0/0
show ipv6 route
ping 2001:db8:1:1::2

show ipv6 interface brief lists each interface with its link-local address and its global address. show ipv6 route shows IPv6 routes, where connected networks are labelled C and local addresses are labelled L.