Where to buy Internet IP addresses

Joe Greco jgreco at ns.sol.net
Sun May 3 05:01:10 CDT 2009


> Mikael Abrahamsson wrote:
> > On Sat, 2 May 2009, Matthew Palmer wrote:
> 
> >> [1] Just because we've got a lot of it, doesn't mean we should be 
> >> pissing it
> >> up against the wall unnecessarily.  A motto for network engineers and
> >> economists alike.
> > 
> > You can't be wasteful with something that you know is already extremely 
> > plentyful.
> 
> There is no resource that cannot be pissed away, if you try hard enough.

Oh!  You mean, like the way we piss away IPv4 addresses?

> > We currently have 72 million billion /56:es. If we do /56:es of the 
> > current /16 being handed out and then change our mind, we can still hand 
> > out 1100 billion /56:es before we can discover this was wasteful and 
> > then we will have spent one 65536th of the address space available.
> 
> Philosophical:
> 
> You know, if we are just going to act as if ipv6 has only 64 bits, why 
> didnt we just design it with only 64?

Well, if you look at it the right way, we *did*.  While you can drill down
into the lower 64 bits, the design intent was clearly to let that be
handled by end users and stateless autoconfig, etc.  So, from a design 
viewpoint, you treat a /64 the way we treat an IP address.  Except that
unlike an IP address, there's a lot more /64's, and each /64 can hold an
IPv4 Internet - squared.

> Practical:
> 
> Think of the routing table. How many /48's in a /32?
> 
> 
> > Give people a /56 and if they only use one NOW,
> 
> Give people a /96. Thats a whole internet.

No!  That's very bad.  That breaks stateless autoconfig.  That's awful.
What in the world is the justification for that?  We can give every person
likely to be on the planet in the next 100 years a /48 and not be even
remotely close to running out.

> Whats that? They resurrected classfull addressing for ipv6?

Pfft.

We *want* things like IPv6 stateless autoconfig to work.  It's a great
idea.  We *want* a protocol simple enough that we don't have to deal
with stateful DHCP, we *want* something that is hard to screw up.

We also don't want to force everyone into a single broadcast domain, or
to have to break stateless autoconfig by subnetting deeper than a /64. 
Sure, *today*, the average person has a handful of networked devices, and
*today*, the average person might not need more than a "public" net and a
"local/printer" net.  However...  twenty years ago, only a few of us had 
a network at home.  It's hard to predict what will be coming in twenty 
years, but I expect there'll be more networking, not less.
Microprocessors are getting smaller and cheaper.  The devices that support
Internet connections are growing, such as all the ethernet ports that have
sprouted on televisions in recent years.  In order to get multiple
broadcast domains and retain stateless autoconfig, you give people more
space than a /64.

But there are plenty of reasons that you might want the wifi on a separate
network from the home computer network, and both of that separate from 
other networks with highly restrictive access, such as a network for the
printer, or your light switches, or your HVAC management, or your alarm
system, or your gaming network, or your kitchen electronics, or your A/V,
etc.  Under IPv6, there will (hopefully) be a demand for home networking
switches and firewalls that greatly simplify security, and one of the
things I'd like to see is the separation of devices at L3, so that a home
user can use a graphical firewall tool that simply determines which
networks are accessible from which other networks in a visual manner,
etc.  This is so much easier with something greater than a /64.

So if you can justify giving people a /96, but you can't see why it'd
be better to give them (at least!) a /56 instead, that just makes no
damn sense.  IPv6 was built around the concept that we wanted to make
space as close to *free* as possible.  The designers intended that it
not be a horribly big deal to have to choose between a /64 and a /56 
and a /48; space is plentiful enough that if you're not absolutely
positive that one size is adequate, that you should then go a little
larger.  Just to be safe.

Because, remember, we're talking tiny slivers of space.  Think about
what a /48 *means*.  Today, you get a single IP address on the IPv4
Internet, one single IP address, and that's a /32.  We're talking much
smaller allocations, and a /48 is an extremely powerful allocation from
the user's point of view.

We will find creative uses for the space.  Look at RFC3041, which is a
credible use for a /64's worth of space that provides a variety of
benefits. 

Any IPv4 engineer who has been working for the last decade will look at
IPv6 and be aghast over the apparent waste, but take a little time to
look at the realities and possibilities before reaching rash conclusions. 
I do see the sheer vastness of a /64 as something that's very difficult 
to wrap your head around: all of the space on today's Internet would be 
just a drop in a /64 bucket.  However, if you can get past that and see
the /64 as just a basic "network," that's a good first step in
understanding IPv6.  Then you should be thinking about the fact that we
will probably be living with this for many years to come, and if we make
poor choices now about how we allocate to users, that will haunt us for
years to come.

... JG
-- 
Joe Greco - sol.net Network Services - Milwaukee, WI - http://www.sol.net
"We call it the 'one bite at the apple' rule. Give me one chance [and] then I
won't contact you again." - Direct Marketing Ass'n position on e-mail spam(CNN)
With 24 million small businesses in the US alone, that's way too many apples.




More information about the NANOG mailing list