[arin-ppml] Equality in address space assignment
Adam Thompson
athompso at athompso.net
Wed Apr 15 20:51:14 EDT 2015
There's a difference between "number of hosts" and "number of public IP
addresses needed".
I've had a number of clients in the 1k-10k IP hosts category who only
needed a handful of, portable, public IPv4 addresses; everything else is
internal use or NAT'd.
The solution isn't too hard, you can artificially use up ~127 IP
addresses pretty easily, and avoid (technically) lying to ARIN, but it's
silly.
But for companies that aren't willing to play games like that (and there
are quite a few, surprisingly), yeah, the "you must be THIS tall" thing
is a problem.
The related issue is when connecting to an IX, you need an AS. To get
an AS, you need PI. To get PI, you need XXX utilization of an
*already-ISP-delegated non-portable* /24. So effectively, ARIN *also*
shuts out smaller companies from connecting to local IXs like MBIX
(where I'm on the board). Yes, it can be worked around, but it's
another hoop you have to jump through, and, usually in my experience, I
see companies simply lying to ARIN to get around it.
(And, yes, I'm well aware of private ASNs - those are simply not allowed
at most IXPs, including MBIX.)
I've been complaining about this since I realized that multi-homing was
no longer sufficient justification to get IPv4 PI, but AS USUAL most
people here didn't care about the [V]SMBs. Or maybe I didn't communicate
clearly, but certainly no-one suggested that. In fact, my comments
effectively vanished into a black hole, as expected.
*I HAVE LESS THAN 61 HOSTS*. Even including VMs and hosted customers.
Even including all the subnets, I would have been fine with a /25 and I
could even have made do with a /26. (Admittedly, I didn't know that at
the time.) And I'm multi-homed, and connected to the local IX. The
ONLY reason I needed PI space was to successfully advertise it using BGP
to multiple providers over a mix of public and private peering.
And despite being the smallest of small networks, my - redundant -
routers can handle roughly 40+ copies of the full routing table right
now. So when I see companies the size of MCI, Google, Comcast, etc.
whining about router memory exhaustion? I bit the bullet and sized my
gear for multi-homed, multi-carrier IPv4+IPv6 including a hefty margin
(like, 500%) for growth. Yes, I'm only running one set of border
routers, not 1000, but that - to me - is an merely argument against the
gigantic networks the US - and this mailing list - is full of.
The lower bar I'd suggest? Anything. No bar at all. Let IPv4 run out
completely, and then, maybe, my local ISPs will finally start thinking
about supporting IPv6.
Rather pissed off with this conversation and everyone who thinks small
networks don't exist or don't matter,
-Adam Thompson
On 2015-04-15 07:28 PM, Jason Schiller wrote:
> On Tue, Apr 14, 2015 at 5:34 PM, David
> Huberman<David.Huberman at microsoft.com
> <mailto:David.Huberman at microsoft.com>>wrote:
>
> So I ask:
>
> How is RIPE and APNIC’s policy unfair, but ARIN’s policy of “you
> must be THIS large a network to participate” fair?
>
>
> When you say "you must be THIS large a network to participate" you are
> talking about networks that are smaller than 61 hosts (plus a router,
> network address and broadcast address), and also don't have a plan to
> have a total of 123 hosts in one year.
>
> (the numbers go down a bit if you have a reason to need more subnets).
>
> I ask for all posters that reference this policy being unfair to small
> networks, to disclose if they have less than 61 hosts, or cannot meet
> a plan for 123 hosts.
>
> This bar was intended to prevent anyone who wanted their own address
> space from getting it and routing it, and contributing to the global
> routing table.
>
> Routing table size IS a real concern, even with current hardware.
>
> (and it is not just 572K IPv4 routes and the 24K IPv6 which cost
> between 1.9 and 2.2 times as much space, but also all of the internal
> routes that large networks are nice enough to aggregate for everyone,
> and the number of routes as a function of your architecture and
> peering locations)
>
> How much lower of a bar would you suggest?
>
> Striking the 61 hosts now and leaving a promise of 123 or more in a
> year's time?
>
> Would you want to lower it 29 hosts now or 13 hosts?
>
>
> ___Jason
>
> On Tue, Apr 14, 2015 at 5:34 PM, David Huberman
> <David.Huberman at microsoft.com <mailto:David.Huberman at microsoft.com>>
> wrote:
>
> RIPE and APNIC policy HAVE ALWAYS ALLOWED all first-time
> requestors to be an LIR and get address space directly from the
> registry. In fact, for more than a decade, that size was
> defaulted to a /20. It is smaller now (a /22 I believe). Then
> to get more space, you show you efficiently utilized what you have.
>
> At the mics at ARIN 35 just today, the large ISPs and Cablecos got
> up to the mic and said a policy which allows small networks to get
> a /24 just by asking for it is unfair.
>
> So I ask:
>
> How is RIPE and APNIC’s policy unfair, but ARIN’s policy of “you
> must be THIS large a network to participate” fair?
>
> What is the technical basis for not allowing small networks to get
> PI space?
>
> Decades of RIPE and APNIC policy didn’t break the internet.
>
> David
>
> *David R Huberman*
> Principal, Global IP Addressing
>
> Microsoft Corporation
>
>
> _______________________________________________
> PPML
> You are receiving this message because you are subscribed to
> the ARIN Public Policy Mailing List (ARIN-PPML at arin.net
> <mailto:ARIN-PPML at arin.net>).
> Unsubscribe or manage your mailing list subscription at:
> http://lists.arin.net/mailman/listinfo/arin-ppml
> Please contact info at arin.net <mailto:info at arin.net> if you
> experience any issues.
>
>
>
>
> --
> _______________________________________________________
> Jason Schiller|NetOps|jschiller at google.com
> <mailto:jschiller at google.com>|571-266-0006
>
>
>
> _______________________________________________
> PPML
> You are receiving this message because you are subscribed to
> the ARIN Public Policy Mailing List (ARIN-PPML at arin.net).
> Unsubscribe or manage your mailing list subscription at:
> http://lists.arin.net/mailman/listinfo/arin-ppml
> Please contact info at arin.net if you experience any issues.
--
-Adam Thompson
athompso at athompso.net
+1 (204) 291-7950 - cell
+1 (204) 489-6515 - fax
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://lists.arin.net/pipermail/arin-ppml/attachments/20150415/d99c3647/attachment.htm>
More information about the ARIN-PPML
mailing list