[ppml] those pesky users...

Stephen Sprunk stephen at sprunk.org
Tue Mar 27 19:16:56 EDT 2007


First of all let me make a request of you: please rate-limit your postings 
to PPML.  It is not necessary to reply to every single message, and repeated 
postings, to the tune of several per hour, is going to make even people who 
agree with you tune the whole discussion out.  Wait a few hours, consolidate 
your responses to similar messages into a single reply, and let others have 
their share of the (very large) audience's limited time.

Now, on to my one response of the day...

> From: "Owen DeLong" <owen at delong.com>
>>Short of a reserved v6 prefix which maps to v4 numbering, this
>>doesn't make sense to me.

There is such a prefix, i.e. ::/96.  However, it's pretty much useless for 
addressing packets since both ends would need be IPv6-capable hosts with 
IPv4 addresses, in which case they might as well use v4 and save a few 
header bytes.

Thus spake "Ted Mittelstaedt" <tedm at ipinc.net>
> Not now, but in the future IF we switch to IPv6 then everyone WILL
> need IPv6 who has IPv4 assigned.

Um, duh?

IPv4-only and IPv6-only hosts can't talk to each other.  The transition 
model, which has been well-documented for over a decade now, has _always_ 
been that everyone goes to dual-stack, and when that's done everyone goes 
back and turns off IPv4.  There has never been a plan in place to go 
directly from one to the other without dual-stacking.

( One might argue that's a flaw, possibly a fatal one.  However, that is 
definitely off-topic here and should be taken up with the IETF. )

> Exactly, that is why this wavier on IPv6 is a bad idea.  Every year it
> is going to get renewed, in the name of "encouraging" people to
> switch to IPv6, and every year more and more IPv6-only freeloaders
> will be out there.

The IPv6 fee waiver requires that orgs either have IPv4 allocations, in 
which case they're paying substantial fees already, or they're a general 
member, in which case they're paying $500/yr for that status.  There will be 
no freeloaders.

If a significant number of IPv6-only networks appear, which has not happened 
yet, the waiver could be modified or not extended.  For instance, I doubt 
many would object to changing the waiver such that members only have to pay 
either their IPv4 fees or their IPv6 fees, whichever is greater.  There's no 
point making that change today, though, since it'd have the same effect as 
the existing waiver.

Note that the cost to ARIN for IPv6 initial allocations/assignments is 
negligible if the applicant is already a member in good standing.  It's not 
like IPv4 (today) where staff must spend time requesting and reviewing 
justifications, utilization, etc. for every request and LIRs must make 
incremental requests every few months, in turn causing staff to repeat that 
effort over and over and over.

It's not until an org needs a second block (or enlargement of their existing 
block) that those hassles kick in for IPv6, and it should be a long, long 
time (if ever) before most folks need that kind of space.  And, since ARIN's 
fees are a direct result of the amount of work needed to provide services, 
resources that require less work to register naturally mean lower fees. 
This is not an accident; policies were steered to this result by design.

> By the time that we have reached the point that there is more IPv6
> assigned than IPv4, there will be heavy political pressure to not get
> rid of the wavier.

ARIN members have to make sure that ARIN meets its operating costs; if the 
members decide IPv6 should remain free as long as IPv4 fees can support 
operations, so be it.  If IPv4 really does go away, the waiver cannot be 
sustained and the members will be forced to drop it.  If they don't, as you 
suggest will happen, the BoT is obligated to override them in order to 
maintain financial viability.  I don't think we, as a community, are so 
stupid as to let that happen, though.

> It's better now to just get rid of it. I believe there is sufficient
> proof that monetary incentives have not encouraged much growth in
> IPv6 assignments.  Just assign everyone IPv6 and be done with it.

So you're simultaneously proposing that ARIN allocate/assign resources to 
people who haven't requested them _and_ that ARIN then bill people for 
having those resources?  I really doubt that combination would pass muster 
with legal counsel.

You can't have it both ways.

> Excellent analysis.  So, given all the above, what ratio do you think
> it appropriate?

We already have policies on how much space to give people who ask for it.  I 
suggest you go read the NRPM and the IRPEP docs, as many of your arguments 
show a general lack of understanding of existing policy as well as how ARIN 

> If you want it to go to IPv6 then your going to have to come to grasp
> with the FACT that IF the Internet goes to IPv6 that eventually

Some will have assignments, some will have allocations; I don't think anyone 
has ever debated the necessity of having one or the other, though.  We all 
grasp it just fine.

> You are saying it isn't feasable now to assign all IPv4 holders IPv6.
> Well if it isn't feasible now, it won't be feasible in the future, either.

Of course it's feasible.  But does it make sense?  I say no.  Existing 
policy makes it absolutely trivial for anyone with an existing IPv4 
allocation or assignment to get their first IPv6 allocation or assignment. 
All they have to do is want one enough to fill out a form.  The fact is that 
even with zero cost and effectively infinite supply, there is simply no 
demand worth speaking about.

Fix the demand first before you start worrying about policies and fees, as 
we've already done everything possible within ARIN's processes to make those 
obstacles nonexistent (to the point we have factions arguing that things are 
_too_ easy and need to be tightened).


Stephen Sprunk      "Those people who think they know everything
CCIE #3723         are a great annoyance to those of us who do."
K5SSS                                             --Isaac Asimov 

More information about the ARIN-PPML mailing list