[arin-ppml] An article of interest to the community....

Joe St Sauver joe at oregon.uoregon.edu
Wed Aug 31 16:41:33 EDT 2011

John asked for thoughts on the scenario:

#   Hypothetical Small College received a /16 from ARIN in order
#   to put all their offices and dorms on the Internet in 1998.
#   Recently, they've realized that they probably requested too 
#   much space long ago, and have also not used in efficiently 
#   (due to reserving significant fixed sized address blocks for 
#   each and every building on campus.)  The good news is they've 
#   figured  out how to overlap multiple address blocks on single 
#   network segments (which they had to learn one the first huge
#   building appeared), and have a plan which would let them redo
#   the entire campus to fit in single /19 address block, or maybe
#   an /18 total (with another /19 reserved for long-term growth. 

Distinguish two case:

-- the campus runs NAT/PAT, and uses private addresses internally
-- the campus uses public IP addresses throughout

I will assume that we're *not* talking about the first of those two
scenarios. In the case of sites using public IP addresses throughout, 
I'd argue that even a /19 could go pretty quickly, even for a 
relatively small institution (like a prototypical liberal arts college
with a few thousand students). 

The old days, when users had one device per person, are gone. These
days it is far more common to see multiple devices: laptop, smart 
phone, tablet, game console, TV digital VCR device, VoIP phone
device, etc., etc., etc. Some of those devices may be NAT'd by the
user, or run on private airgapped networks, but at least some of those 
devices may also move around. Giving them addresses via DHCP with 
moderately persistent lease times can result in even a single device 
(like a laptop) potentially tying up multiple addresses across multiple
subnets, at least for periods of time when the user is moving around. 

Sites also need to anticipate and cope with shifting peaking user loads.
For example, residence halls might be largely empty when classes are 
in session, while classroom buildings with large lecture halls might
need many IPs during that period of time; in the evening, the reverse
may be true. Nonetheless, DHCP pools need to be sized to accomodate
those peaking loads (and dynamically resizing those DHCP pools would 
likely be an exercise in folly, as would attempting to run an 
unsegmented "flat" network architecture). 

I'd also note that as sites move to smaller blocks, their ability to 
multihome and get those smaller aggregates announced and accepted will 
decrease. That may or may not also be a consideration for some sites. 

#   Is their space unused?  

For all the reasons described above, no, their space is not unused (or 
even "underutilized") in many cases (in my opinion). 

#   They can't possibly free it up 
#   without first doing a significant amount of effort, but 
#   they have every reason to list some portion of their IP
#   block for auction (as long as they are clear about the
#   timing on delivery.)

I don't think it's correct to say that they have "every reason to list
some portion of their IP address block for auction."

I'd characterize that as an ultimately self-defeating "auto-canibalistic"
"eating of one's seed corn" strategy, potentially prematurely disposing of 
an irreplaceable core asset in a way that's kin to selling "excess" real 
estate at a campus that is surrounded on all sides by other growing and 
long term tenants. Once an asset of that sort is gone, you're never going 
to be able to get it back, and by disposing of that asset you've just 
capped your ability to grow. Yes, you got some cash, but you paid for it 
in flexibility and long term options, and those may be major "expenses."

I'm delighted when people are altruistic, and do their best to return
community assets that they don't need, but I'm not surprised when people
act in economically rational ("selfish") ways that maximize their own
benefit to the potential detriment of the community.

For those users, the economically rational folks who ARE thinking about 
selling resources (rather than simply returning unneeded resources to 
ARIN for reuse), selling an asset such as "excess" IPv4 address space 
now, before the market has fully exhausted available resources and thus 
before market prices have had a chance to reach well 
understood/predictable levels, strikes me as a potentially unsound choice.
Put another way, "timing's everything" and I'm not sure now's the right
time if you're a wanna-be IPv4 address seller or broker. 

While it's possible that IPv6 will suddenly become universally deployed, 
and the market value of IPv4 addresses will slump, I don't think that's
a very likely scenario. I think that it's more likely that IPv4 address 
space will continue to become more scarce, continue to be in demand, 
and thus IPv4 addresses will continue to appreciate in value.

If ARIN wants to incent liquidity in the IPv4 address market space, I 
think it needs to address that (perceived) reality. That would imply 
convincing people that new sources of IPv4 address space will come on 
line soon, so that those who are currently holding excess assets (with 
an eye towared longer-term appreciation) decide to "move" those assets 
now, rather than waiting. 

While we all know that Class E IPv4 addresses cannot currently be 
usefully deployed, if there was movement to try to correct that, that's
the sort of thing that might incent some who are currently "holding 
onto" IPv4 address assets to move their "inventory" while there's
still a potential resale market. (And yes, I know, bringing on Class 
E addresses might not buy much more than eighteen months of breathing 
room, but sometimes even "small" perturbations can have large 
*perceptual* impacts to systems as complex as the potential resale 
market for IPv4 address space).

Alternatively, consider how cities make sure that real estate gets 
highest and best use: taxes. You could grow soybeans in Manhattan,
but if you tried, the taxes would kill you. Taxes are a potentially
important tool for signaling that soybeans work better in sparsely 
populated great plains, and skyscrapers are a higher and better use 
of limited real estate in the city. 

Currently, the "tax" imposed on IPv4 address space per year 
(https://www.arin.net/fees/fee_schedule.html) is arguably too low 
to motivate a resource holder to dispose of assets that aren't 
needed, and therefore aren't being put to highest-and-best 
productive use. Complicating that, at least some "assets" are 
"tax exempt," anyhow. :-;

All in all, it's a fascinating economic system and set of questions
to consider, I think.



Disclaimer: all of the above is just my POV, and you'd have to be crazy
to rely on these thoughts for anything with potential economic or 
operational impacts. :-)

More information about the ARIN-PPML mailing list