IPv6

IPv6 has 128 bits and three types: unicast, multicast and anycast.  Unicast and multicast act the same way as they do in IPv4, anycast is the new type and allows multiple interfaces to have the same address and works by the nearest interface answering the request.

IPv6 Special Addresses

• Global Unicast - similar to IPv4 publicly routed addresses, start with 2000::/3
• Link-Local- address for one-one communication not routed internally, starts with FE08:/10
• Unique-Local- similar to IPv4 private routed addresses, starts with FC00::/7
• Multicast- similar to IPv4 multicast, starts with FF00::/8
• Local Host- 0:0:0:0:0:0:0:1
• DHCP stateful - 0:0:0:0:0:0:0:0
• IPv6 Mix-mode- 0:0:0:0:0:0:192.168.0.1
• 6 to 4 Tunneling - 2002::/16

Auto configuration

In IPv6, a host machine has the ability to address themselves with a link-local address and a global unicast address. Stateless auto configuration works by contacting the router getting prefix ID, similar to network ID in IPv4 and using its MAC address padded with 16 bytes, eui-64. Stateless auto-configuration does have one drawback, it doesn't automatically get DNS information from the router.

1. The host sends out a router solicitation (RS) message to all routers using a multicast address requesting pre-fix information.
2. The router answers the RS with a router advertisement (RA) including the prefix ID using multicast  as well, RAs are sent out periodically. 

Configuring IPv6 

Router(config)# ipv6 unicast-routing
Router(config-if)#  ipv6 address 2001:db8:3c4d:1::/64 eui-64



ICMPv6 works pretty much the same way it does in IPv6 but it's not a individual layer 3 protocol anymore, it's built into IPv6. My default ICMPv6 prevents fragmentation using MTU discovery. ARP isn't used anymore with IPv6 to discover machines MAC addresses, ICMP neighbor discovery (ND) sends out multicast to host who then reply back with their layer 2 addresses. multicast Listener Discovery has replaced IGMP to notify routers that a host has join a multicast group.

RIPng fundamentally works the same in IPv6 , still uses 15 hops, sends its whole routing table periodically has split-horizon and route poisoning.  Instead of using multicast addresses 224.0.0.9 for updates RIPng uses FF02::9 RIPng is configured at the interface level while additional commands can still be configured at the router level. Words can also be used to ID the router process .

Router(config-if)# ipv6 rip [NUM or WORD ] enable

Router(config)# ipv6 router rip 1


EIGRPv6 works the same as regular EIGRP but is configured differently. EIGRP must be turned on with no shutdown command and configured on each interface instead of using the network command. Instead of using 224.0.0.10 multicast address it uses FF02::A (A=10 in hexadecimal)

Router(config)# ipv6 router eigrp 10
Router(config-rtr)# no shutdown
Router(config-if)# ipv6 eigrp 10


OSPFv3 works the same but is configured different. The RID must be configured manually because RID is a 32 bit field and IPv6 is 128 bits. OSPF is created at the router level along with setting the RID, networks can be added on interfaces. OSPF multicast are sent at FF02::5 (IPv4 224.0.0.5)for OSPF routers and FF02::6 (IPv4 224.0.0.6)for designated routers.


IPv6 Migration 

• Dual Stacking - allows devices to use both IPv4 and IPv6 simultaneously 
• 6to4 Tunneling- allows IPv6 network to communicate across IPv4 networks, teredo allows tunnel to use UDP packets to get past NAT