Sunday, May 29, 2011

Lesson 57 - Introduction to IPv6 - Address Configuration

It's time to put our hands to work now. Let's see how we can configure IPv6 addresses. I'm going to use the following topology diagram to present different options with regard to IPv6 address configuration on router's interfaces.

Pic. 1 - IPv6-IPv4 Topology Diagram.
Icons designed by: Andrzej Szoblik -

I will use R1 and R3 to begin with. The whole 'Introduction to IPv6' series will end with connecting these two IPv6 networks over IPv4 cloud to show you one of many migration options available.

As you already know, each IPv6-enabled interface, be it a router or computer's interface, will use two distinct addresses:
  • link-local (FE80::)
  • global (2000::)
Currently there are two global prefixes in use (as of writing this post):
  • 2001 - global addresses
  • 2002 - 6to4 tunnel (without actually creating a tunnel interface).
Let's open R1's console and configure IPv6 address on F1/0 interface. First, I'll use EUI-64 method to derive the host bits based on the MAC address. Take a look:

Pic. 2 - IPv6 Address with EUI-64.

Note, that by just configuring IPv6 on the interface, IPv6 protocol stack gets turned on automatically. You can verify this like shown below:

Pic. 3. IPv6 Address Verification

There area two methods of turning on IPv6 support on an interface in IOS:
  • By using 'ipv6 enable' command.
  • By configuring global IPv6 address.

Both link-local and global addresses are now configured. The subnet is what I typed in, the host portion created based on the MAC address taken from F1/0 interface. Do we always have to use EUI-64? Of course not. Both link-local and global addresses can be created automatically. I have removed the previous configuration (pic. 3) and configured this instead:

Pic. 4 - IPv6 Manual Address Assignment
(link-local and global).

Both addresses (link-local and global) are configured manually. Verify this below now:

Pic. 5 - Verification.

Okay. Let's bring the R1's interface F1/0 up, and jump over to R3 for a moment to configure its F1/0 interface ... differently!

An alternative to the above methods (it is going to be used primarily on the computers rather than routers) is so called 'autoconfig' method. For this method to work you must configure two commands first (as of writing this post in most IOS versions I came across both of these must be turned on):

R1 Configuration:
R1(config)#ipv6 unicast-routing

R3 Configuration:
R3(config)#ipv6 unicast-routing
R3(config)#int f1/0
R3(config-if)#ipv6 enable
R3(config-if)#ipv6 address autoconfig

Let's verify configuration of IPv6 with this method!

Pic. 6 - Verification.

The moment I enabled 'ipv6 unicast-routing', the router begins to support this method. R3 configured as 'autoconfig' will obtain the network portion of the global address (the most significant 64 bits) from the local router (here: R1) by sending a special message called: Router Solicitation (RS). R1, upon receiving this message is going to reply with Router Advertisement (RA). This way, R3 learns what the network portion of the global address is and using EUI-64 completes the host portion, thus creating full, globally unique address to operate.

In IPv6 world there are no broadcasts, so the neighbor discovery using ARP is no longer working. The neighbor discovery, layer 3 to layer 2 address mapping are done using ICMPv6 protocol. There are four ICMPv6 Neighbor Discovery messages you must be aware of. These are:
  • NS - Neighbor Solicitation, asking for neighbor information.
  • NA - Neighbor Advertisements, introducing yourself to the neighbors.
  • RS - Router Solicitation, asking for info about local routers.
  • RA - Router Advertisements, advertising yourself as an active router.
Let's enable few debugs (remember that these are dangerous commands on production equipment) and see few of these messages being exchanged.

Pic. 7 - Debugging ICMPv6 Communication.

ICMPv6-ND: DAD: FE80::CE02:6FF:FE1E:10 is unique.
The above messages is the Duplicate Address Detection (DAD; mechanism of checking if the link-local is unique).

ICMPv6-ND: Sending NA for FE80::CE02:6FF:FE1E:10 on FastEthernet1/0
IPV6: source FE80::CE02:6FF:FE1E:10 (local)
      dest FF02::1 (FastEthernet1/0)
After checking that R3's link-local address is unique, it introduces itself to all the neighbors on F1/0 interface (Neighbor Advertisement).

ICMPv6-ND: Sending RS on FastEthernet1/0
IPV6: source FE80::CE02:6FF:FE1E:10 (local)
      dest FF02::2 (FastEthernet1/0)
R3 is sending Route Solicitation trying to discover locally active routers in order to obtain the information about the network portion of the global address.

ICMPv6: Received ICMPv6 packet from FE80::1, type 134
ICMPv6-ND: Received RA from FE80::1 on FastEthernet1/0
R3 receives information from R1 (R1's Router Advertisement)

ICMPv6-ND: Sending NS for 2001:13:13:13:CE02:6FF:FE1E:10 on FastEthernet1/0
IPV6: source :: (local)
      dest FF02::1:FF1E:10 (FastEthernet1/0)
R3 is sending the Neighbor Solicitation trying to discover the neighbors. I already knows the 2001:13:13:13::

ICMPv6-ND: Autoconfiguring 2001:13:13:13:CE02:6FF:FE1E:10 on FastEthernet1/0
R3's autoconfiguration is complete.

ICMPv6-ND: DAD: 2001:13:13:13:CE02:6FF:FE1E:10 is unique.
R3's 128-bit global address is unique (Duplicate Address Detection).

ICMPv6-ND: Sending NA for 2001:13:13:13:CE02:6FF:FE1E:10 on FastEthernet1/0
IPV6: source 2001:13:13:13:CE02:6FF:FE1E:10 (local)
      dest FF02::1 (FastEthernet1/0)
      traffic class 224, flow 0x0, len 72+8, prot 58, hops 255, originating
IPv6: Sending on FastEthernet1/0
ICMPv6-ND: Address 2001:13:13:13:CE02:6FF:FE1E:10/64 is up on FastEthernet1/0
R3 is announcing its presence on F1/0 segment (Neighbor Advertisement) and fully configured to operate using IPv6 address.

Note, that the ND (ICMPv6 Neighbor Discovery) communication uses multicast (FF). Two addresses are used here:

FF02::1 - All local hosts on the link.
FF02::2 - All local routers on the link.

One last thought about EUI-64 and link-local address. Since, it is locally significant, this address will be used on ALL interfaces. The reason I mention this is that if you try to check connectivity between link-local interfaces, you must provide the router with the output interface. It uses this address on ALL of them after all. Also, you must type in the full interface name to send icmp echo message using link-local address. Check it out:

Pic. 9 - Ping Connectivity Test.

Let's check R3's neighbors (layer 3 to layer 2 mapping):

Pic. 10 - IPv6 Layer 3 to Layer 2 Mapping.

Yes, this would be your IPv6 equivalent of the 'show arp' command used in IPv4!

In my next post we will configure RIPng (RIP Next Generation) routing protocol. The 'Introduction to IPv6' will finish with a short discussion on IPv4 to IPv6 migration and a quick lab to see how we can route IPv6 packets over IPv4 networks.


CCNA Basics | Cisco Advanced | Linux Last Update: 19/02/2019 - eBGP Peering CCNA Lessons: 01 - Connecting to Cisco Cons...