[arin-ppml] Draft Policy ARIN-2013-4: RIR Principles - revised

David Farmer farmer at umn.edu
Sat Jul 13 00:03:42 EDT 2013

On 7/12/13 12:34 , William Herrin wrote:
> On Thu, Jul 11, 2013 at 7:58 PM, David Farmer <farmer at umn.edu> wrote:
>> We still have IPv6 and ASNs to worry about, and while both
>> resource pools are GARGANTUAN by comparison, they are not infinite.
> Hi David,
> While not technically infinite, the 32 bit ASN pool is functionally
> infinite. There is no foreseeable course of events which would exhaust
> that pool. Ever.

As ASNs are currently used, 1 or a very small number of ASNs per 
organization, they might seem functionally infinite.  However, the IETF 
is about to allocate 94,967,295 Private Use ASNs[1] (AS4,200,000,000 - 
AS4,294,967,294) which is more than 1400 times the ASNs currently in 
use.  This is intended to facilitate other BGP applications like large 
scale enterprise use of BGP based MPLS VPNs, or some new data center 
uses of BGP[2].  Think of an ASN per top of rack or end or row switch in 
your favorite monster scale Data Center.

[1] http://tools.ietf.org/html/draft-ietf-idr-as-private-reservation-05
[2] http://tools.ietf.org/html/draft-lapukhov-bgp-routing-large-dc-04

In the discussion of expanding Private Use ASNs it was suggested it 
might be better if blocks (maybe large blocks) of unique public ASNs 
were assigned to organizations, instead of going the private use route. 
So, if large blocks of ASNs were to be assigned per organization, then 
ASNs might not seem so functionally infinite anymore.

This is all hypothetical right now, except the nearly 100M Private Use 
ASNs, that has been approved by the IESG and is in the RFC Editor Queue 
now.  But, if in the future we start using ASNs in different ways than 
now, and our rate of use changes significantly, then ASNs may not seem 
so functionally infinite any more.  In the mid-80s 4B IP addresses 
seemed functionally infinite.  But, by the mid-90s it was obvious 4B IPs 
wasn't functionally infinite at all.  Is 4B ASNs still going to seem 
functionally infinite in 2025?  I think it will probably still seem 
functionally infinite, but all I know is things change.

>> 1. Can everyone concede that going forward, conservation is much less
>> important, but that the need for some concept of conservation doesn't
>> completely go away either.
> I'd like to, but no. This document applies it to "number resources"
> not "addresses." Previous documents on the subject (e.g. RFC 2050)
> correctly limited conservation statements to "addresses."
> Also, conservation of IPv6 addresses moving forward is neither "much
> less" important nor even "less" important. It's merely different.

"Much less" is in comparison to other issues for IPv6, conservation is 
much less important than routability for IPv6 in my opinion.

> We'd like to not burn through that free pool in the next decade or
> two. It wasn't that long ago that some folks in this forum quite
> seriously suggested allocating /16's and /15's to ISPs in support of
> 6rd efforts.

People say lots of things, I don't think anyone really thought a /16 for 
6rd per ISP was reasonable.  One reason was conservation.

> Unlike the ASN pool, the IPv6 pool is not functionally
> infinite.

I believe some form of conservation is applicable for all "number 
resources", even ASNs.  Again, are ASNs going to seem functionally 
infinite in 2025?  Also, similar to Matthew requesting a /8 of IPv6 
earlier today; If we don't need conservation for ASNs then I'll take a 
million ASNs please.

I've provided what I think is a plausible situation where conservation 
applied to ASNs could be beneficial.  Can you provide a plausible 
situation where conservation applied to ASNs could be harmful?

> Finally, IPv4 conservation beyond the free pool is moot. One cannot
> conserve what has already been consumed. Redistribution, in whatever
> form it takes, is not properly described as conservation.

The definitions I'm thinking of;
Conservation - Noun - The action of conserving something, in particular.
Conserving - Verb - Prevent the wasteful or harmful overuse of (a resource).

A Hypothetical; Is it in the interest of Internet for a bankruptcy 
trustee to sell IPv4 addresses to a snow shoe spammer?  How is something 
like this prevented without a concept of conservation for IPv4? 
Personally, I don't have to use the "C" word, but the interconnected 
meaning of a lot of the concepts we are talking about boil down to the 
same or only subtly different things.


David Farmer               Email: farmer at umn.edu
Office of Information Technology
University of Minnesota
2218 University Ave SE     Phone: 1-612-626-0815
Minneapolis, MN 55414-3029  Cell: 1-612-812-9952

More information about the ARIN-PPML mailing list