What is addressing in IPv6 in detail.
The increase of the address length from 32 bits up to 128 on logic will increase also Routing tables. To not to let it be, it is necessary to build hierarchical system of addressing. An example such systems is telephone system, where on a code of the country, cities and then lines of automatic telephone exchange it is possible to track a route up to the subscriber.
IPv6-address can be introduced in three forms :
1) The basic form: x:x:x:x:x:x:x:x
Here x is 16-bit hex number (i.e. having 4 hex symbols).
2) The compressed form.
Here, for reduction addresses which contains some groups comprising only zero bits, reduction "::" is applied - it means, that on its place there is any quantity of groups with zero bits. This type of record as follows will look:
(1 form) 108b:0:0:0:8:800:200C:417A
(2 form) 108b::8:800:200C:417A
(1 form) ff01:0:0:0:0:0:0:43
(2 form) ff01::43
(1 form) 451a:0000:0000:0000:e63d:0000:0000:8264
(2 form) 451a::e63d:0:0:8264.
Loopback address :
(1 form) 0:0:0:0:0:0:0:1
(2 form) ::1
It is clear, that we cannot replace some groups of zero bytes with a double colon, as in this case it will be not clear, how many the byte is passed in the first, and how many in all other cases.
3) The alternative form.
It is obvious, that transition on IPv6 will be smooth: it's impossible to simultaneously transit all the devices connected to the Network on a new infrastructure. Especially for "vague time" in IPv6 it is stipulated two types of "transitive" addresses. The first reveal to the knots, critical for tunneling of the traffic between IPv6 and IPv4, and they consist of 96 zero bits and customary addresses IPv4. The second are intended to the knots, which do not support the new system of the addressing: 80 zero bits are pluses of 16 single bits, and then to — address IPv4 (RFC 2373).
in compressed form:
In IPv6 there are following types of addresses: unicast, anycast and multicast . Addresses conduct not to node and to the interface of this node because each interface belongs only to one node.
Unicast IPv6 address corresponds only with one interface. To one interface can correspond much IPv6 addresses of various types (unicast, anycast and multicast). There are two exceptions of this rule:
1. The single address can be attributed to several physical interfaces if the application considers these some interfaces as a node at its representation at a Internet level.
2. Routers can have not numbered interfaces (for example interface that haven't its own IPv6 address) for point-to-point connections to exclude necessity manually configuring and declaring (advertise) this addresses. These addresses are not necessary for connections of point-to-point routers if these interfaces isn't using as points of departure or destination at transferring of IPv6 datagrams. In this case routing is carried out under the scheme near to using by CIDR protocol in IPv4.
IPv6 Corresponds to model IPv4 where the subnetwork associates with the channel. To one channel there can correspond some subnetworks.
Type of IPv6 address is defined by its leading bytes. Analyzing these bytes it is possible to understand, to what type address concerns. It simplifies routing. The combination of the leading bytes is called prefix.
The following is the table of prefixes, their purposes and parts of the space borrowed by them :
Apparently, at such distribution it is easy to distinguish addresses of the ISP, local addresses and multicast addresses.
The syntax of unicast address commonly contains prefix, index of network area (Area ID), subnetwork ID, other rote indexes and in the end - interface ID which is it's unique IEEE-802 MAC address.
Anycast addresses are part of unicast address space and their syntax is the same as unicast addressees. Such address can't identify a PC. It can only be a router's address and it cannot be specified in field "Source Address".
Syntax of such address consists of subnetwork prefix and others null-bits i.e. it's just like unicast address of interface with zero identifier. The packages sent to such address, will come to all routers of a network, but work will be really carried out with the one which will send the first answer.
Multicast IPv6 addresses have following structure:
In field Flags3 older bits are reserved and the last (T) means: T = 0 address is standard ("well-known") multicast.
T = 1 address is temporary ("transient").
Field Scope is a 4-bit code of multicasting, intended to defining of limited scope. Admissible values:
Listed above multicast addresses are reserved and will not be appropriated any multicast groups.
Addresses for the reference to all nodes:
Listed above addresses identifies group includes all IPv6 nodes within group1 (local nodes) or group2 (local connected nodes).
Addresses of all the routers :
Listed above multicast addresses identifies group of all IPv6 routers within area1 (local nodes) or area2 (local connected nodes).
DHCP server/relay-agent: FF02:0:0:0:0:0:0:C
Listed above multicast addresses identifies group of all IPv6 DHCP servers and transit agents within area (scope) 2 (local channel.
Listed above multicast address is counted as function of unicast and anycast addresses of this node. Solicited-node consists of younger 32bits of address (unicast or anycast) adding 96bits prefix FF02:0:0:0:0:1. The result is multicast address coveres interval :
For example solicited node's multicast address code corresponding IPv6 address 4037::01:800:200E:8C6C is FF02::1:200E:8C6C. IPv6 addresses which differs only older bits will coincide with solicited node's address.
Based on material by Semenov U.А. - "IPv6 Addressing" (Original link).