[AusNOG] IPv6 rudiments (was Re: IPv4 - Free /48 from APNIC)

[Karl Auer]

On Tue, 2013-03-05 at 15:23 +1300, Don Gould wrote:
> I don't fully understand this concept of prefixes.  192.168.1.1 where's 
> the prefix in that?  192?

Prefixes are the same in IPv6 as they are in IPv4 CIDR, just with more
bits. A prefix consists of an address and a prefix length. The prefix
length says how many bits, counting from the left, belong to the network
part and how many bits are left for subnetting and host addresses
("Interface Identifiers" in IPv6-speak) within the network. The notation
is very similar for IPv4 and IPv6:

   192.168.1.0/24 means a 24-bit IPv4 network
   2001:db8::/32 means a 64-bit IPv6 network

Without the prefix length, you can't tell what the prefix is. Is your
example an address in 192.0.0.0/8, 192.168.0.0/16 or 192.168.1.0/24? Or
perhaps it's an address in any of the many other possible prefixes, like
192.8.0.0/9, or 192.168.1.0/26... but if you write an address with a
prefix length, you should make sure that the context makes clear that it
is an address, not a prefix. Easy to get confused otherwise.

> Currently I think in terms of a single address per device.  Are you 
> saying all my devices get two addresses and then just use one?

IPv6 interfaces will generally have *at least* one address, *usually*
three or four, and *commonly* five to ten (more than four is less
common). Common candidates:

   IPv4 address (DHCPv4)
   Link local address (SLAAC)
   Global Unicast Address (SLAAC)
   Global Unicast Address (DHCPv6)
   Global Unicast Address (SLAAC privacy preferred)
   Global Unicast Address (SLAAC privacy deprecated)
   Global Unicast Address (SLAAC ULA)
   Global Unicast Address (SLAAC ULA privacy preferred)
   Global Unicast Address (SLAAC ULA privacy deprecated)

> I don't understand how the devices know which is the least cost route.

They don't, any more than an IPv4 device does, unless they run a routing
protocol. Absent a routing protocol, there is only a first-hop
algorithm. In a multi-router scenario, the routers can offer rudimentary
guidance. See RFC4861 and RFC 4191, or for a simplified version see my
blog post here: http://into6.com.au/?p=229.

Routing in IPv6 (except for first hop routing) works the same as in
IPv4. No new things to learn, though obviously the nuts and bolts of the
actual communication are different (OSPFv3 uses link-local addresses,
for example). First hop routing in a single-router scenario is pretty
simple.

> Clearly I need to make a list of all the skills I didn't know I needed 
> just to have enough learning to replace my v4 functionality. I wonder 
> what else I don't know that I don't know I need to know?

I think you may be suffering from analysis paralysis. Just do it; you'll
be amazed how easy most of it is. Sure there are a lot of subtleties,
but basically it works like IPv4. There are a few areas where you need
to change your thinking, but 90% of everything stays the same or very
similar.

BTW, you will not be replacing your IPv4 functionality - you'll be
extending it. Dual stack will be the way to go for some time to come.

Regards, K.

-- 
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
Karl Auer (kauer at biplane.com.au)
http://www.biplane.com.au/kauer
http://www.biplane.com.au/blog

GPG fingerprint: B862 FB15 FE96 4961 BC62 1A40 6239 1208 9865 5F9A
Old fingerprint: AE1D 4868 6420 AD9A A698 5251 1699 7B78 4EEE 6017