[clug] IPv6 Australia?

Robert Edwards bob at cs.anu.edu.au
Thu Jul 24 23:40:03 GMT 2008

Sam Couter wrote:
> Robert Edwards <bob at cs.anu.edu.au> wrote:
>> When the IPv4 address crunch comes, all that will happen is that a
>> market will open up for IPv4 address space. Those large corporations
>> etc. sitting on several class A's will make some money breaking them
>> up and flogging off the unused addresses as class C's.
> Excellent. Yet more money to be made by rich people by selling an
> artificially scarce resource.

That would be the rich people who invested lots of time and staff salary
dollars in developing the Internet in the first place...

When a disruptive technology comes along, the playing field changes.
Some people make some money, others may have to spend some. Get over it.

>> Unlike Sam (and many others), I think that NAT is cool and don't hate
>> it at all. All properly designed protocols work fine with NAT, so why
>> not?
> NAT relegates IP devices to the role of client only. They cannot act as
> a server. IP is supposed to be peer based. NAT breaks that.

IP back in the 60s when all hosts belonged to well managed organisations
and users accessed networks via multi-user time-sharing hosts (cf. a NAT
router/server with several/many "client-only" desktops/laptops/PDAs etc.
sitting behind it) was peer-to-peer.

> The biggest downside of the peer relationship being widely broken is
> that it becomes difficult to publish a service that a big player isn't
> willing to host. Imagine how far everyone's favourite and most abused
> protocol, HTTP, would have got if the big players at the time like AOL
> were in control and weren't interested in supporting a competitor to
> their already established distribution networks.

I have a single IPv4 address at home. I have a single web-server
sitting on/behind that address. I can (and do) "publish" several
websites on that address. Anything serious goes to a web-hosting
company (ala the "big players"?). I suspect that most people would
operate this way, even if they did have 2^64 IPv6 addresses at home.

I can also connect into my single IPv4 address at home over SSH (or
some other protocols) and manipulate all manner of devices that exist
in my house but are not visible on the rest of the Internet. Some of
those devices have a very small CPU (and carbon foot-print) and I
don't want to have to set up all sorts of firewall rules on them or
on their behalf on my stateful firewall. They have absolutely no need
to be visible from the rest of the Internet and rarely need to connect
out. If my future fridge manufacturer wants to ping my fridge to check
how cold it is (or maybe spy on what products I keep in it), then I
would prefer them to come in via an authenticated service on my IPv4
address/NAT router/server, not directly to my (future) fridge.

>> So no compelling advantage for IPv6 other than more address space (to
>> defend against in your firewall scripts/blacklists etc.) and bigger
>> addresses (takes more CPU to hash when connection tracking and more
>> memory to store etc.). DNS for IPv6 is a real doozy (esp. reverse DNS!).
> On the other hand, the strict hierarchical routing scheme means routers
> don't need to store such massive routing tables.

We'll wait and see how "strictly" adhered to the hierarchical routing
scheme really ends up. I am not so sure. The so-called "massive routing
tables" of IPv4 aren't all that massive these days, with CIDR and other
policies having reduced the potential number of entries. In any case,
it is not an issue for most people and those who do care have solved
it or are using IPv6 as their backbone routing infrastructure.

IPv4 will eventually go away. Whether it is replaced with IPv6 or some
other much more exciting protocol, we'll wait and see. I suspect that
a new "killer-app" protocol will come along before IPv6 is widely
adopted and we'll end up jumping over it (cf. FDDI back in the 90s).


Bob Edwards.

More information about the linux mailing list