[arin-ppml] Policy Proposal 102: Reduce and Simplify IPv4 Initial Allocations

Ted Mittelstaedt tedm at ipinc.net
Mon Nov 9 18:14:36 EST 2009


Seth Mattinen wrote:
> William Herrin wrote:
>> On Mon, Nov 9, 2009 at 3:48 PM, Ted Mittelstaedt <tedm at ipinc.net> wrote:
>>> However, multihomers who qualify for a /22 right now are completely
>>> free to KEEP that /22 and go get another /22, when requesting an
>>> additional allocation.  Thus, creating 2 DFZ entries.
>>>
>>> Under this proposal we put a stop to that.  Multihomers no longer exist
>>> as a special class, when ANYONE gets a /22 and then goes and requests
>>> an additional allocation, they will have to return that first /22 to get
>>> their /21.  Thus, knocking then back down to 1 DFZ entry.
>> Hmm. I missed that on the first read. Not sure what I think of it.
>>
> 
> That's what I was opposing, not availability to small entities.
> 

Seth,

   Well, then, here's the crux of the issue.

   When a small entity obtains multiple /22's under section 4.2.2.2, they
create multiple advertisements in the DFZ, in Bill's words - costing
all of the rest of us more money.  Bill's assertion is that small 
entities costing us all money is unfair, which is why he opposes 
modification to 4.2.2.1.   Bill refuses to discuss his opinion on the
CURRENT 4.2.2.2, but clearly he must oppose the current minimums in
4.2.2.2 also - or he would be inconsistent - which is why when I
pointed out that the policy proposal gets rid of that loophole (for
smal multihomers) he says now that he no longer has an opinion.

   In short, he's being consistent - his basic position is anything that
increases the DFZ is bad.

   You, by contrast, are making a different argument.  Your opposition 
isn't based on what is or isn't good for the DFZ, your basic position - 
clarify me if I'm wrong - is that once ARIN hands out a sub-/20
allocation, the receiving org should be able to regard it as a permanent 
assignment.  If you were being consistent then you would NOT oppose this
proposal if it only affected orgs under section 4.2.2.1  (ie: the
single-homers)

   The problem I see (and I am guessing that Chris sees as well) is that
there's no way that someone coming from Bill's position of anti-DFZ
growth could ever be satisfied with a proposal that made someone from 
your position, (the permanent assignment position) happy.  These 2 
positions are incompatible.

   If we rewrite a proposal that only affects 4.2.2.1  (for example, 
suppose I modified Proposal 98 to get rid of the triggers and just
make it apply immediately) so that the permanent assignment people
are happy, then the anti-DFZ-growth people are going
to oppose it because it would increase the DFZ, because the multihomer
loophole will still be around.

   If we rewrite Section 4.2.2.2 to include forced return-and-renumber,
(essentially what this proposal does) then the anti-DFZ-growth people
will be happy but the permanent assignment people will not be.

   The anti-DFZ-growth people are already unhappy about the DFZ loophole 
in 4.2.2.2 as it is.  And the small single-homers who are currently
tied to LIR-assignments that they are afraid will go away post-IPv4
runout, are EXTREMELY unhappy about things as it is now - and would
be overjoyed to get anything at all - like this proposal.

   So, like Chris, I have to basically say that the most we could do
is possibly respond to your criticism of it by playing around with
the math a bit - like for example a renumbering requirement of /22
like Chris was suggesting - but even then we risk opposition from
the anti-DFZ-growth people, so if playing around with the math isn't
enough for the permanent assignment people to change their opposition to
support, then it would be pointless to bother trying even that.

   As I see it, the small sub-/20 multihomer ISPs out there have had
plenty of time to get their initial allocations under Section 4.2.2.2.
They know IPv4-runout is coming, the bar was made lower for them to
get their portable IPv4 a long time ago.  If there's any of them out 
there who absolutely must get their "never-will-need-to-renumber-/22"
then since this policy proposal change is going to take a half a year to 
implement, they are going to have PLENTY of warning.  And, in 2-3 years 
it won't matter anyway, since there won't be virgin IPv4 at the RIR's
left.

   As much as I'd like to accommodate the permanent assignment people, I 
think there's a lot more of the anti-DFZ-growth people out there, and in 
this proposal I'm frankly more interested in looking out for the small 
single-homers who I really believe are going to be jacked around once 
the IPv4 transfer market starts up.

Ted



More information about the ARIN-PPML mailing list