[182275] in North American Network Operators' Group
Re: Dual stack IPv6 for IPv4 depletion
daemon@ATHENA.MIT.EDU (Owen DeLong)
Tue Jul 14 16:19:10 2015
X-Original-To: nanog@nanog.org
From: Owen DeLong <owen@delong.com>
In-Reply-To: <0C521EF9-B55B-4F3A-9E9F-F4F5B918C467@beckman.org>
Date: Tue, 14 Jul 2015 12:56:58 -0700
To: Mel Beckman <mel@beckman.org>
Cc: "nanog@nanog.org" <nanog@nanog.org>
Errors-To: nanog-bounces@nanog.org
I expect to be actively involved at least 20 more years. If I'm not around,=
thats 160 years to runout. I'm betting the protocol can't live that long fo=
r other reasons.=20
Owen
> On Jul 14, 2015, at 12:35, Mel Beckman <mel@beckman.org> wrote:
>=20
> I have no problems with ISPs giving out /48s to residential subscribers. N=
either do I mind if they give out /56s. That still gives every residential c=
ustomer 256 /64 subnets.=20
>=20
> I don't see this as something that needs to become a standard. Those end-u=
sers who want more can ask for more fro their ISP whenever the need arises. I=
f there is a market for selling those larger prefixes to end users, that's f=
ree enterprise, which I also support.=20
>=20
> I don't think it's wise to delegate by rule or convention that the entire f=
irst 1/8th of IPv6 space should be delegated in /48s. You see this as not a h=
uge deal. To me, 12.5% is a huge deal.
>=20
> I appreciate your offer to give your services away for free to remedy any p=
roblems the /3 bolus creates. But as history has shown, neither of us is lik=
ely to be in circulation -- or even alive -- when a problem would occur. =20=
>=20
> -mel beckman
>=20
>> On Jul 14, 2015, at 10:30 AM, Owen DeLong <owen@delong.com> wrote:
>>=20
>> You don=A1=AFt think holding nearly 7/8ths of the address space in reserv=
e for a future addressing policy is adequate judiciuosness?
>>=20
>> The IPv4 /8s constituted 1/2 of the address space. The /16s another 1/4, a=
nd the /24s an additional 1/8th at the time.
>> Overall, that was 7/8ths of the address space assigned to unicast and 9/1=
6ths allocated if you included multicast.
>>=20
>> In IPv6, we have 1/8th set aside for unicast, 1/256th for multicast, 1/25=
6th for ULA, 1/1024th for link-local, and
>> a couple of infinitesimal fractions set aside for other things like local=
host, IPV6_ADDR_ANY, etc.
>>=20
>> As I said, let=A1=AFs be liberal as designed with the first /3. If I=A1=AF=
m wrong and you can prove it in my remaining lifetime, I will happily
>> help you develop more restrictive allocation policy for the remaining 3/4=
while the second /3 is used to continue growing the
>> IPv6 internet.
>>=20
>> Whatever unexpected thing causes us to finish off the first /3 likely won=
=A1=AFt burn through the second /3 before we can
>> respond with new policy. We still have almost 3/4 of the address space av=
ailable for more restrictive allocations.
>>=20
>> Frankly, I bet about 1/8th of the IPv4 address space probably is in the h=
ands of the top 64 organizations. Maybe more.
>>=20
>> In this case, 1/8th of the address space will more than cover the entire k=
nown need many many many times over, even
>> with very liberal allocations.
>>=20
>> Owen
>>=20
>>> On Jul 14, 2015, at 10:13 , Mel Beckman <mel@beckman.org> wrote:
>>>=20
>>> Owen,
>>>=20
>>> By the same token, who 30 years ago would have said there was anything w=
rong with giving single companies very liberal /8 allocations? Companies tha=
t for the most part wasted that space, leading to a faster exhaustion of IPv=
4 addresses. History cuts both ways.=20
>>>=20
>>> I think it's reasonable to be at least somewhat judicious with our spank=
ing new IPv6 pool. That's not IPv4-think. That's just reasonable caution.=20=
>>>=20
>>> We can always be more generous later.=20
>>>=20
>>> -mel beckman
>>>=20
>>>> On Jul 14, 2015, at 10:04 AM, Owen DeLong <owen@delong.com> wrote:
>>>>=20
>>>> 30 years ago, if you=A1=AFd told anyone that EVERYONE would be using th=
e internet 30 years
>>>> ago, they would have looked at you like you were stark raving mad.
>>>>=20
>>>> If you asked anyone 30 years ago =A1=B0will 4 billion internet addresse=
s be enough if everyone
>>>> ends up using the internet?=A1=B1, they all would have told you =A1=B0n=
o way.=A1=B1.
>>>>=20
>>>> I will again repeat=A1=A6 Let=A1=AFs try liberal allocations until we u=
se up the first /3. I bet we don=A1=AFt
>>>> finish that before we hit other scaling limits of IPv6.
>>>>=20
>>>> If I=A1=AFm wrong and we burn through the first /3 while I am still ali=
ve, I will happily help you
>>>> get more restrictive policy for the remaining 3/4 of the IPv6 address s=
pace while we
>>>> continue to burn through the second /3 as the policy is developed.
>>>>=20
>>>> Owen
>>>>=20
>>>>=20
>>>>> On Jul 14, 2015, at 06:23 , George Metz <george.metz@gmail.com> wrote:=
>>>>>=20
>>>>> That's all well and good Owen, and the math is compelling, but 30 year=
s ago if you'd told anyone that we'd go through all four billion IPv4 addres=
ses in anyone's lifetime, they'd have looked at you like you were stark ravi=
ng mad. That's what's really got most of the people who want (dare I say mor=
e sane?) more restrictive allocations to be the default concerned; 30 years a=
go the math for how long IPv4 would last would have been compelling as well,=
which is why we have the entire Class E block just unusable and large block=
s of IP address space that people were handed for no particular reason than i=
t sounded like a good idea at the time.
>>>>>=20
>>>>> It's always easier to be prudent from the get-go than it is to rein in=
the insanity at a later date. Just because we can't imagine a world where I=
Pv6 depletion is possible doesn't mean it can't exist, and exist far sooner t=
han one might expect.
>>>>>=20
>>>>> On Tue, Jul 14, 2015 at 12:22 AM, Owen DeLong <owen@delong.com <mailto=
:owen@delong.com>> wrote:
>>>>> How so?
>>>>>=20
>>>>> There are 8192 /16s in the current /3.
>>>>>=20
>>>>> ISPs with that many pops at 5,000,000 end-sites per POP, even assuming=
32 end-sites per person
>>>>> can=A1=AFt really be all that many=A1=A6
>>>>>=20
>>>>>=20
>>>>> 25 POPS at 5,000,000 end-sites each is 125,000,000 end-sites per ISP.
>>>>>=20
>>>>> 7,000,000,000 * 32 =3D 224,000,000,000 / 125,000,000 =3D 1,792 total /=
16s consumed.
>>>>>=20
>>>>> Really, if we burn through all 8,192 of them in less than 50 years and=
I=A1=AFm still alive
>>>>> when we do, I=A1=AFll help you promote more restrictive policy to be e=
nacted while we
>>>>> burn through the second /3. That=A1=AFll still leave us 75% of the add=
ress space to work
>>>>> with on that new policy.
>>>>>=20
>>>>> If you want to look at places where IPv6 is really getting wasted, let=
=A1=AFs talk about
>>>>> an entire /9 reserved without an RFC to make it usable or it=A1=AFs pa=
rtner /9 with an
>>>>> RFC to make it mostly useless, but popular among those few remaining N=
AT
>>>>> fanboys. Together that constitutes 1/256th of the address space cast o=
ff to
>>>>> waste.
>>>>>=20
>>>>> Yeah, I=A1=AFm not too worried about the ISPs that can legitimately ju=
stify a /16.
>>>>>=20
>>>>> Owen
>>>>>=20
>>>>>> On Jul 13, 2015, at 16:16 , Joe Maimon <jmaimon@ttec.com <mailto:jmai=
mon@ttec.com>> wrote:
>>>>>>=20
>>>>>>=20
>>>>>>=20
>>>>>> Owen DeLong wrote:
>>>>>>> JimBob=A1=AFs ISP can apply to ARIN for a /16
>>>>>>=20
>>>>>> Like I said, very possibly not a good thing for the address space.
>>=20
