[arin-ppml] ARIN-prop-167 Removal of Renumbering Requirement for Small Multihomers

[Owen DeLong]

On May 3, 2012, at 2:07 AM, Ted Mittelstaedt wrote:

> On 5/2/2012 11:58 PM, Owen DeLong wrote:
>>> Yes it's difficult to renumber.  But, the org renumbering is
>>> getting something for their trouble - that is, they are getting
>>> more IP addresses.  Many small end user orgs in the past have
>>> renumbered-and-returned just fine under 4.3.6.2  I don't see that
>>> suddenly in year 2012 that something new and special has come along
>>> that now makes renumbering impossible for these orgs.
>>> 
>> 
>> Uh, no, Ted, they haven't. Please refer back to Leslie's policy
>> experience report.
>> 
>>> But ARIN must put a barrier up to simply request-without-renumber
>>> otherwise the end user orgs will simply not do it.  The proposal
>>> is
>> 
>> Why is that so bad? Today, an organization that needs a /20 is free
>> to go out and purchase 16 discreet /24s and advertise all of them.
> 
> yeah, yeah please stop with the 8.3 gong banging.
> 

Either disaggregation is a concern and we should, therefore focus on the
largest sources of such disaggregation, or, it isn't.

Pretending that this proposal is bad because of the potential for disaggregation
which is rather mathematically limited while accepting the disaggregation
consequences of 8.3 which are virtually unlimited seems disingenuous at
best and an effort to lock out the little guy to subsidize larger players at worst.

>> We're talking about an organization that started with a /24 or a /23
>> expanding to a maximum of 4 /24s before the policy becomes moot
>> anyway.
> 
> I would submit that most orgs that start with a /24 either are low-to-no
> growth and will never really need anything beyond the /24, or they
> have a good enough mousetrap that people are clawing for what they
> have and they can easily justify jumping from a /24 to a /22 on the
> first go around.
> 

In which case, this proposal would have even less impact than what
I suggested.

I was trying to place bounds on the worst case.

>> In fact, the current way for an organization that has a /24
>> and doesn't want to renumber to get to having 2 /24s worth of space
>> is quite simple. Entity A creates corporation B and multihomes
>> corporation B. Entity A moves half of their need into corporation B
>> to qualify for an ARIN /24, then moves corporation B back inside
>> Entity A through section 8.2.
>> 
> 
> More gong banging.
> 

I'm not sure why you call this gong banging. Bottom line, it's a very
simple way to circumvent the policy we are seeking to repeal, so,
the policy is merely an inconvenience, doesn't achieve the desired
objective, and is far from being the worst offender among existing
policies for the so-called problem at hand.

> Do you know how much it costs today in the US to setup a corporation?
> 

About $60 if you incorporate in Delaware, but, you don't have to incorporate
to be an organization in the ARIN region. There are many other valid
structures which are even cheaper.

> If it's worth it that much to the small org to keep it's /24 and not
> renumber, then go for it!  They will make some lawyer happy and
> pay for the down payment on his yacht.  Then a year later when they
> want to use a /23 mask internally they will have screwed themselves.

$60 to avoid the pain of renumbering? Sign me up. Heck, I pay $100/year
in ARIN fees to avoid the pain of renumbering and that's just for my house.

> Meanwhile their competitors will spend the money on a better network infrastructure and be done with their renumbering.

Really? How much network infrastructure can you buy for $60?

> I seem to remember you always advising this approach when it came
> to shifting from IPv4 to IPv6, Owen.

Apples to oranges comparison. If you can call gong-banging (whatever
that means), I think this qualifies as hand-waving.

>>> throwing the baby out with the bathwater and has no recognition
>>> for the benefit to the community of forcing orgs to use contiguous
>>> subnets.
>>> 
>> 
>> But there is no baby and the bathwater appears to smell pretty foul
>> at this point.
>> 
>> Org's today aren't forced to use contiguous subnets. With 8.3, that's
>> only going to get a whole lot worse going forward.
>> 
> 
> Yeah well I get that your POed about 8.3 - write a policy proposal to
> fix it, please, and stop trying to make every other policy proposal
> discussion a referendum on 8.3

I'm actually not POed about 8.3. I do think it's bad policy and I had my
say about that. The community spoke and if you review the record, you'll
find that the arbitrary deletion of the sunset clause was my only reason
for voting against the policy. You'll also find that I voted in favor of 2011-1
and 2012-1 updates to 8.3.

While I am not a fan of 8.3, I accepted that the majority of the community
supported it and certain aspects of it were unfortunately vital for the
time being. My subsequent comments about 8.3 have, I believe,
been limited to:

1.	Discussions of proposals to modify 8.2/8.3
2.	This proposal

Basic arithmetic says that 8.3 has much larger routing table implications
than this proposal. My 8.3 comments aren't about trying to make this a
referendum on 8.3. They're an effort to get the community to recognize
the hypocrisy inherent in accepting 8.3 as currently written while objecting
to this proposal.

Owen
(Speaking solely for myself and not in my capacity as an AC member)