Re: [address-policy-wg] Is the time for conservation over?

[Sascha Lenz <]

Hay,

Michael.Dillon@localhost wrote:
> One area of address policy that is fairly consistent world-wide is the
> view that IPv4 address space is scarce and that the policy must
> be conservative, i.e. the policy must make conservation of IPv4
> addresses a high priority.
>
> I don't think that's true anymore. 
Right, i share this point of view for quite a while now, and also 
recently made some side-notes on this in some mails to the 
address-policy-wg list, too.

My main motivation to change my opinion on this from "conserve!" to
"waste!" was, that there is IPv6 now, and obviously noone found _any_ 
killer-application for IPv6 yet. So not many ISP managers see any reason 
to invest any time or money in IPv6, and almost no End-User does even 
know about IPv6.
The only real "killer reason" for IPv6 is, that it's just needed if 
there is no more IPv4 space left.
So, it's a bit populistic, but "waste IPv4 space, now!" is in general 
what should be done :-)

Secondly, I also always thought that this whole "there is no more IPv4 
left soon!" thing couldn't be true based on the very slow development 
nowerdays and the huge reserves. But i never mentioned it since i didn't 
really do any research about that and had no facts.
If i look at the recent documents and talks about this (like the one 
during RIPE meeting mentioned earlier 
http://www.potaroo.net/ispcolumn/2003-07-v4-address-lifetime/ale.html 
for example), i'm quite sure now that this is a second reason why we 
should rethink the conservation policy.

But some more annotations on your opinion:

> On the one hand, we have IPv6 deployed commercially in 3 of the
> 4 policy regions (Europe, AsiaPac, America) which indicates a continuing
> trend toward a future time where IPv6 service will be almost as 
> easy to find as IPv4 service. On the other hand, the worst possible 
> outcomes discussed when CIDR was first deployed are not going
> to happen. For instance there was a fear that the People's Republic
> of China might want 1/4 of the IPv4 space because they have 1/4
> of the planet's population. This has not happened and is now quite
> unlikely to happen.
There's the main problem, I'm in .de - quite many LIRs with IPv6 
Allocations here... but, can i get native IPv6 conenctivity anywhere?
As End-User almost not at all, as ISP for some Uplink? Same.
(Yes, i know there are some ISPs, but it's far from "as easy to find as 
IPv4 Uplinks" yet).

So i don't really see this coming anytime soon. As long as there is IPv4 
space left, i really doubt that descent IPv6 deployment will happen.

> Therefore, I believe that all the RIRs should jointly do some research
> to establish a prudent date at which IPv6 will be considered to have
> reached critical mass so that there will be a significant migration of
> users from IPv4 to IPv6. Once we set our sights on this date we should
> set aside a certain amount of buffer in the IPv4 space, and then design 
> our policy to consume the rest of the IPv4 space, not to preserve it.
> At the same time, this policy shift should be presented as part of
> a global IPv6 migration strategy because that is what it is. 
I don't really get what you want this for.
Don't make things more complicated - make them easier!
I.e. there is some special verificaton policy on static dialin IPs ect. 
in the RIPE Assigment documents.
Or the "no reserverations allowed" paragraph is also a bit outdated 
since there are some new things like Sub-Allocations introduced now.

==> I only know the RIPE Policys in detail, but when talking about
    re-thinking the convervation policy, i mainly talk about
    re-thinking all the special-policies, just strike them out probably.


> In addition, I don't see any good reason to wait until LIRs come and ask
> for IPv6 space. It's not scarce and the vast majority of IPv4 LIRs will
> be deploying IPv6 sometime. So why don't we just give every single
> one of them an IPv6 /32 today. Instead of creating barriers to the 
> adoption
> of dual v4/v6 networks as we are today, we should be facilitating the
> operation of dual v4/v6 networks. We need to create an environment
> in which the end user can choose whether to use v4 or v6 rather than
> constraining the end users with our v4-centric regulatory bureaucracy.
Whereas i strongly support a change of the IPv6 Allocation Policy to
"give it to any LIR who wants it; strike out any requirements" for
the reasons you gave here, the problem still remains - most IPv6 
projects are more or less "private fun" of some network operators, very 
few ISPs do support it officially or have any plans for supporting IPv6 
natively in their whole network any time soon (yes, there are positive 
exceptions of course).

And there should be a formal request of course, no need to waste any 
IPv6 block if some LIR doesn't really want to have one for whatever reason.
But just kick out the requirements we have as "slow start" mechanism up 
to now.
I also had some nice discussions with some RIPE hostmasters and on the 
mailinglists about why the hell they require a network diagram in the 
IPv6 (Allocation) request form... probably someone remembers my rants 
(btw, i won) :-)

--
========================================================================
= Sascha Lenz                  SLZ-RIPE          slz@localhost         =
= Network Operations                                                   =
= BayCIX GmbH, Landshut                  * PGP public Key on demand *  =
========================================================================