Re: [address-policy-wg] 2007-08 New Policy Proposal (Enabling Methods for Reallocation of IPv4 Resources)

[Iljitsch van Beijnum iljitsch@localhost]

On 31 okt 2007, at 2:08, David Conrad wrote:

> > > > About half the ~ 40 legacy /8 assignments don't show up in the
> > > > routing table.
> > > Which, of course, means precisely nothing.
> > The value of an IP address is the ability to receive packets from
> > elsewhere addressed to it. Without a presence in a routing table
> > someplace, that doesn't happen so the IP address is of no value.
> The fact that _you_ can't see a routing announcement for a  
> particular prefix
> does NOT mean the prefix isn't announced somewhere.
True. But then do you concede to the basic logic so the only question  
is which routing table we look at?
> There are these things
> called "private networks" and they do interconnect outside of the  
> context of
> the "public" Internet.
Sure, but is it reasonable for (for instance) the US government to  
have several percent of the total IPv4 address space and use it for  
this when we have space set aside for exactly these purposes? Japan is  
the second country in the world as far as IPv4 address use goes with  
140 million addresses (but not for long, China is going insanely fast)  
which is a bit lower than what the US government holds. Not the US -  
that would be 1.4 billion addresses out of the 2.55 billion in use -  
but just the US GOVERNMENT: around 10 /8s.
> I make the assumption
> that markets are going to exist regardless of whether folks stamp  
> their feet
> and pout about their existence.
If we can't do anything about it, why are we having these discussions?

> The fact that long prefixes will
> undoubtedly be made available could have potential negative  
> implications, no
> question.
I'm not that worried. That's only going to be an issue when an ISP  
that now gets their space in /12 blocks needs to take 256 /20s. I  
don't think those ISPs will be prepared to pay market price for that  
much space, NATing customers will be cheaper for them.
> However, it would seem best to try to address (pun intended) that
> issue directly instead of pointlessly trying to address it  
> indirectly by
> commanding the tide to not come in.
Saying there will be a market is harmful regardless of whether it's  
true, because that way, people will be disinclined to give back the  
address space they currently hold but don't use. So if there's going  
to be one, let it be a surprise.
> > It requires herculian effort to keep the up-and-coming
> > economies happy with the way the internet is currently "run" (if  
> > there
> > is such a thing).
> Actually, it doesn't.  Your view is somewhat condescending.  Folks in
> developing countries are as involved in the way the Internet is  
> currently
> "run" (in terms of setting address policy) and are as aware of the  
> issues as
> are folks in developed countries.
Read up on the positions of the Chinese and Brazilians (or rather,  
their governments) on "internet governance". When was that whole  
circus again? Last year, the one before?
> > What is the developing world going to say when they
> > have to pay rich American companies for address space--address space
> > that those companies got for free?
> They will be unhappy.  Perhaps a bit less unhappy than being told  
> "it is
> impossible to obtain any additional address space, period", but  
> perhaps not.
I think it's better for poor countries if we're all out, that way it's  
everyone's problem, not just one for those who can't afford the  
remaining scraps. Misery loves company.
> The reality is that the IPv4 address space is running out and as  
> long as
> there is continued demand for IPv4 address space, there are going to  
> be
> people who are able to obtain address space and some who will not,
> regardless of the mechanisms of redistribution.
What needs to happen is that for someone doing network planning, it's  
a better choice to go with IPv6 rather than to fledge the IPv4 horse  
some more. Any and all time we spend making life post-runout easier is  
a waste of time and harmful because it only delays the real solution.  
People are going to do what they're going to do; it's not our jobs to  
make it easier for them to make shortsighted decisions.
> > There are STILL people that refuse to bother implementing IPv6 in
> > their products,
> And WHY are they not implementing IPv6?  Because there is no customer
> demand.
No, it's because they can get away with it. People like Cisco are  
still selling routers with no 32-bit AS support even though this is a  
minor update that they've had around for years and we KNOW we'll need  
this 14 months from now.
> Why is there no customer demand? Because IPv6 provides no technical
> incentive over IPv4.  Since there are no technical incentives, it  
> would seem
> the next best option is financial incentives.  What is your  
> alternative?
If there are no incentives there is no reason to do anything. Hence  
the current situation. No more IPv4 addresses will be an incentive  
soon enough. What needs to happen is that vendors prepare for that and  
don't _wait_ until customers have trouble.Î