Appendix 6 - An Introduction To Ipv6; Vastly Increased Address Space; Standard Subnet Size; Address Allocation - ADDER AdderLink Digital iPEPS User Manual

Appendix 6 - An introduction to IPv6

During the initial design of the Internet, 4.3 billion seemed like an impossibly
large number of device addresses, possibly more than would ever be needed.
It took nearly forty years, but finally the last remaining vacant address blocks
within the current Internet Protocol scheme (called IPv4) were assigned in
February 2011.
The Internet Protocol is a crucial element of Internet operation and the eventual
exhaustion of unique addresses was predicted and acted upon many years
ago. The replacement for IPv4 is known as IPv6 and was defined in December
1998. Since then its uptake has been slow (reportedly used for less than 1% of
Internet traffic in 2008) although this will increase rapidly as places within the
incumbent system are exhausted.

Vastly increased address space

The most notable feature of IPv6 is the size of its address space, put simply: It's
massive. By using 128 bits to define each IPv6 address (rather than the 32 bits
used in IPv4), there are now 340 x 10
trillion trillion or as it is correctly known, 340 undecillion).
The larger address size of IPv6 requires a different manner of notation. Instead
of the four decimal numbers separated by dots used for IPv4 (e.g. 192.168.0.1),
IPv6 addresses consist of eight groups of four hexadecimal digits that are
separated by colons (e.g. 2002:00a2:67be:0000:0000:0e82:8723:a144) –
each group of four digits represents 16 bits of the address. By necessity, IPv6
addresses are quite long and so there are a couple of techniques to help reduce
this in certain cases:
• Where a group has one or more leading zeroes, these can be omitted. In the
above example 00a2 and 0e82 can be written a2 and e82, respectively.
• Where one or more consecutive groups consist solely of zeroes, they can
be replaced with a double colon (::). In the above example, the fourth and
fifth groups could be replaced with the double colon, so that the whole line
could be reduced to: 2002:a2:67be::e82:8723:a144. It is easy to return any
such shortened address to the full version by replacing the double colons
with sufficient groups of zeroes until the total number of groups is returned
to eight. For this to work it is essential that only one set of consecutive zero
groups within an address are replaced with a double colon.
36 unique addresses (that's 340 trillion

Standard subnet size

Thanks to the new huge address space, IPv6 does not need to wring every
last drop out of each address range and so it handles address allocation in a
different manner than its predecessor. Whereas IPv4 uses subnets of varying
sizes (using the Subnet Mask entry to define the size of each subnet), IPv6
subnets are (almost) all set to a standard size. A full 64 bits are used to define
each subnet, which means that every standard IPv6 subnet has use of an address
space that is the square of the entire IPv4 address space (that's 1.8 x 10
addresses per subnet). In those subnets, all addresses are valid host locations;
gone are special address formats for particular uses, such as broadcast traffic.
Also, now that all standard subnets are the same size, the subnet mask is
another item that is made redundant under IPv6.

Address allocation

Every device attached to an IPv6 network usually has more than one address
type. The two most common types are called a link-local address and a global
address and these can be assigned in a number of ways.
In IPv4, device addresses are most commonly assigned either manually or by
using a Dynamic Host Configuration Protocol server (DHCP). IPv6 also offers
manual addressing and DHCP (now called DHCPv6 and fully supported by the
Digital iPEPS unit), but also allows devices to automatically configure their own
addresses using a series of steps defined as StateLess Address AutoConfiguration
(or SLAAC). The key parts of the SLAAC procedure occur roughly as follows:
• The IPv6 compliant device creates a tentative local identifier which is usually
derived from its fixed unique hardware identifier (or MAC address). The
local identifier is 64 bits in length (the lower half of the full 128 bit address)
and this is one of the advantages of having a fixed subnet size; it is very
straightforward to automatically figure out the boundaries and contents of
the local network. This is exactly what the device does next with its tentative
local identifier.
• The device uses the Neighbor Discovery Protocol (part of the Internet Control
Message Protocol suite – IMCPv6) to check within the local network whether
its tentative local identifier is being used by any other device. If it is, then
the device will create a new one and start the process again. If the local
identifier is unique within the local network, it is then combined with the
standard link-local prefix (fe80::) to form a valid link-local address. At this
stage the address is valid only for communication within the local network.
The next stage is to replace the link-local prefix with a global prefix and
then carry out a similar procedure in order to prevent address duplication,
resulting in a validated global address.
19
continued
48