IPv6 Addresses

IPv6 Addresses
So#a Silva Berenguer sofia@lacnic.net Addresses
l 
An IPv4 address is formed by 32 bits.
232 = 4.294.967.296
l 
An IPv6 address is formed by 128 bits.
2128 = 340.282.366.920.938.463.463.374.607.431.768.211.456
~ 5,6x1028 IP addresses for each human being.
~ 7,9x1028 times the number of IPv4 addresses.
Addresses
The IPv6 addresses representation divides an address in 8 groups of
16 bits, separated by “:”, represented by hexadecimal digits.
2001:0DB8:AD1F:25E2:CADE:CAFE:F0CA:84C1
2 bytes
When representing an IPv6 address it is allowed:
l  To use lowercase or uppercase letters
l  Omit leading zeros and ...
l  Represent groups of contiguous zeros by “::”.
Example: 2001:0DB8:0000:0000:130F:0000:0000:140B
2001:db8:0:0:130f::140b
Not valid format: 2001:db8::130f::140b (generates ambiguity)‫‏‬
Addresses
l 
l 
Prefix representation
Like CIDR (IPv4)‫‏‬
l  “IPv6 address/prefix size”
l 
Example:
l  Prefix 2001:db8:3003:2::/64
l  Global Prefix 2001:db8::/32
l  Subnet ID 3003:2
•  URL
l 
l 
http://[2001:12ff:0:4::22]/index.html
http://[2001:12ff:0:4::22]:8080
Addresses
Three types of addresses have been defined for IPv6:
l 
Unicast → Individual Identification
l 
Anycast → Selective Identification
l 
Multicast → Group Identification
Broadcast addresses don’t exist anymore.
Addresses
Unicast
l 
Global Unicast
n
Global Routing Prefix
64 - n
Subnet ID
64
Interface identifier
2000::/3
• 
Globally routable (similar to public IPv4 addresses)
• 
13% of the total of possible addresses;
• 
2(45) = 35.184.372.088.832 different /48 networks.
Addresses
Unicast
l 
Link local
FE80
0
Interface identifier
l 
FE80::/64
l 
It shall only be used locally;
l 
Automatic set-up (Stateless autoconfiguration);
Addresses
Unicast
•  Interface identifier (IID)‫‏‬
§  Must be unique in the same subnet prefix
§  The same IID can be used for different interfaces of the same
node, only if they are associated to different subnets.
§  Usually a 64-bit IID is used This IID can be obtained:
§  Manually
§  Stateless autoconfiguration
§  DHCPv6 (stateful)‫‏‬
§  From a public key (CGA)‫‏‬
§  The IID may be temporary and randomly generated
§  Usually it’s based on the interface’s MAC address (EUI-64
format).
Addresses
Unicast
l 
MAC Address
EUI-64
EUI-64 Address
0
1
0
0
1
0
0
0
1
0
48
48
1E
1E
C9
C9
21
FF
FE
FF
FE
85
0C
21
85
0C
21
85
0C
Bit U/L
0
1
0
0
1
0
Interface ID
Link Local Address:
4A
1E
C9
FE80::4A1E:C9FF:FE21:850C
Addresses
Unicast
l 
Unique local
7
Pref.
L
Global identifier
Subnet
ID
Interface ID
FC00::/7
l 
l 
l 
Globally unique prefix (with high probability of being unique);
It’s used only for communications on a link or on a limited set of
links;
It’s not expected to be routed in the Internet.
Addresses
Unicast
l 
Special addresses
l 
l 
l 
l 
l 
Localhost - ::1/128 (0:0:0:0:0:0:0:1)‫‏‬
Unspecified - ::/128 (0:0:0:0:0:0:0:0)‫‏‬
IPv4-mapped - ::FFFF:wxyz
Special ranges
l  6to4 - 2002::/16
l  Documentation - 2001:db8::/32
l  Teredo - 2001:0000::/32
Obsoletes
l 
Site local - FEC0::/10
IPv4-compatible - ::wxyz
l 
6Bone - 3FFE::/16 (Test network deactivated on 06/06/06)‫‏‬
l 
Addresses Anycast
l 
Identifies a group of interfaces
l 
l 
The packet is delivered only to the interface that is nearest the
source
Anycast addresses are syntactically equal to unicast addresses.
They are used differently.
Possible uses
l 
To discover services in the network (DNS, proxy HTTP, etc.)‫‏‬
l 
For load balancing
l 
To locate routers that provide access to a certain subnet
l 
In networks supporting IPv6 mobility, anycast addresses are used
to locate Home Agents
Addresses
Multicast
l 
Identify a group of interfaces.
l 
Multicast support is mandatory in all the IPv6 nodes.
l 
Multicast addresses come from the block FF00::/8.
l 
The prefix FF is followed by 4 bits used as flags and 4 more bits
that define the scope of the multicast address. The remaining 112
bits are used to identify the multicast group.
8
FF
4
4
Flags
0RPT Scope
112
Multicast group identifier
Mul8cast Address
Scope
Description
FF01::1
FF01::2
FF02::1
FF02::2
FF02::5
FF02::6
FF02::9
FF02::D
FF02::1:2
FF02::1:FFXX:XXXX
FF05::2
FF05::1:3
FF05::1:4
FF0X::101
Interface All the interfaces(all-nodes)
Interface All the routers (all-routers)
Link
All the nodes (all-nodes)
Link
All the routers (all-routers)
Link
OSPF Routers
Link
Designated OSPF Routers
Link
RIP Routers
Link
PIM Routers
Link
DHCP Agents
Link
Solicited-node
Site
All the routers (all-routers)
Site
DHCP servers at a site
Site
DHCP agents at a site
Diverse NTP (Network Time Protocol)
Addresses
l 
l 
l 
l 
As in IPv4, IPv6 addresses are assigned to the physical interfaces
and not to the nodes.
With IPv6 it’s possible to assign multiple addresses to a single
interface, independently from it’s type.
Thus a node can be identified by any address of it’s interfaces.
l  Loopback
::1
l  Link Local
FE80:....
l  Unique local
FD07:...
l  Global
2001:....
RFC 3484 determines the algorithm used to choose the source and
destination addresses.
Resource Allocation and
Distribution Policy
Allocation and Distribution Policy
l 
Each RIR received from IANA a /12 block
l 
The block 2800::/12 is the space reserved for LACNIC
l 
l 
Bigger distributions could be done if the current utilization is
justified
ATENTION! Unlike what happens in IPv4, in IPv6 the utilization is
measured considering the number of address blocks assigned to
end users and not the number of addresses assigned to end users.
Recommendations
Recommendations for the assignment of addresses (RFC3177)
•  In general /48 networks are recommended for all the different types
of users: domestic users, small or big companies.
•  Big companies can get a /47 block or several /48 blocks
•  If there is certainty that only one subnet is needed, /64 networks are
recommended, i.e., for 3G users.
•  A /128 network may be used when there is absolut certainty that only
one interface will be connected.
Providers
l 
l 
NTT Communications
l  Japan
l  Native IPv6 (ADSL)
l  /48 to end users
l  http://www.ntt.com/business_e/service/category/nw_ipv6.html
Internode
l  Australia
l  IPv6 nativo (ADSL)
l  /64 dinámico para sesiones PPP
l  Delega /60 fijos
l  h:p://ipv6.internode.on.net/configura8on/adsl-­‐faq-­‐guide/ Proveedores
IIJ
l 
l 
l 
l 
l 
l 
Japón
Túneles
/48 a usuarios finales
http://www.iij.ad. jp/en/service/IPv6/index.html
Arcnet6
l  Malasia
l  IPv6 nativo (ADSL) o túneles
l  /48 a usuarios finales
l  se pueden distribuir bloques /40 y /44 (sujeto a aprobación)
l  http://arcnet6.net.my/how.html
Consideraciones
/32
•  65 thousand /48 networks (33 thousand if waste is considered)
•  16 million /56 networks (6 millones, si consideramos hd ratio)
es suficiente para su proveedor?
•  Reservar un bloque (/48 ?) para infraestrutura…
Links punto a punto:
/64? /112? /120? /126? /127?
RFC 3531 HERRAMIENTAS
22
Gestión de direcciones IPv6
l 
l 
El tamaño de las nuevas direcciones hace mas engorrosa su
manipulación en forma directa
Veremos el uso de dos herramientas para implementar un caso de
estudio simple
l 
IPPlan
- 
l 
Implementación de la numeración a alto nivel
SIPCalc
- 
Implementación a nivel detallado para un punto de
presencia
IPPlan
IPPlan
l 
l 
l 
l 
l 
IPPlan es una herramienta open source muy conocida para la
gestión de espacio IP
La versión 6 en adelante soporta IPv6
Se puede bajar desde http://iptrack.sourceforge.net
Algunas características:
l 
Interfaz web
l 
Capacidad de importar tablas de enrutamiento
Requisitos:
l 
Apache + PHP (4 o 5) + MySQL
IPPlan
l 
Paso 1: Crear un “cliente” o “sistema autónomo”
IPPlan
l 
Crear un rango de direcciones IPv6 asociado al cliente / sistema
autónomo
l 
“Create a new network area”
IPPlan
l 
Crear subnets de acuerdo al plan de numeración
l 
En nuestro ejemplo dividimos 2001:db8::/32 en 16 subredes /36
IPPlan
l 
Visualizar las subredes para trabajar sobre ellas
SIPCalc
SIPCalc
l 
l 
l 
SIPCalc es una herramienta de línea de comando que permite
trabajar con direcciones IPv6 y realizar algunas tareas comunes
Se puede bajar de: http://www.routemeister.net/projects/sipcalc/
También esta en los repositorios de las distribuciones de Linux/
Unix mas comunes:
l 
l 
l 
Debian / Ubuntu
Fedora / CentOS
MacPorts
SIPCalc
l 
Dividiendo el /36 en dos /37:
SIPCalc
l 
Los primeros 5 clientes empresariales (asumiendo /52 por cliente)
SIPCalc
l 
Los primeros 6 clientes residenciales (asumiendo /56 por cliente)
SIPCalc – DNS Reverso
l 
SIPCalc puede utilizarse para generar reversos de DNS
Comentarios finales
l 
IPPlan y SIPCalc comparten la mayoría de sus funcionalidades.
l 
Ambos son útiles en diferentes escenarios
l 
l 
IPPlan es una herramienta de gestión y de planificación
SIPCalc es una herramienta muy importante para los
administradores de redes en su trabajo diario