[Chapter-delegates] scarcity of IPv4 addresses

Fri Oct 24 11:35:33 PDT 2008

I am not sure how I missed the conversation over the last six weeks and I
apologize for jumping into this "late" but I think that you may want to
conduct further evaluation before making a public stand.  I have been
intimately involved in this discussion here in the ARIN region for over six
months now and I am definitely not convinced that creating a market is the
answer to IPv4 scarcity.  In fact there are several very intelligent and
cluefull individuals who believe exactly the opposite, that an address
market will cause the collapse of the current RIR infrastructure due to
government intervention in many countries which could in turn splinter the
Internet with regionalized or nationalized government regulations that may
not be compatible across borders.  This is not an issue that I believe you
have a high likelihood of being on the correct side of after 6 weeks of
light conversation, IMHO.

If members of this list are interested, I can compile a fairly long list of
reference material and documentation of the views of folks on both sides of
this issue.  To start with, this decision tree may be helpful to some who
have not spent as much time thinking about this issue yet as others
have: http://odin.chrisgrundemann.com/files/Do_I_Support_A_Liberalized_Transfer_Policy.jpg.
<http://odin.chrisgrundemann.com/files/Do_I_Support_A_Liberalized_Transfer_Policy.jpg>

For now I have added some quick comments to the points below.

On Thu, Oct 23, 2008 at 8:59 AM, John Schnizlein <schnizlein at isoc.org>wrote:

> Discussion of this situation and how ISOC should stand on it has continued
> over the roughly six weeks since I asked for discussion of it here.  Thanks
> to those who replied.
>
> Please comment on the following draft ISOC position.
>
> John
>
> 1) ISOC believes there is no practical way to prevent address
> transfers - so we advocate registering the transfers.
>
> Some member of Regional Internet Registries (RIRs) oppose any transfer
> other than recovery of unused addresses to the RIR for re-allocation.  They
> oppose any opportunity for address holders to profit from transfers as
> unfair.  Anecdotes that (part of) companies have already been traded in
> order to acquire address blocks they hold suggest that this opposition is
> futile.  Scarce resources become valuable, and will be traded, either openly
> or secretly.

A) The argument that "it is going to happen anyway" is a very week one
imho.  Under this logic there should be no regulation at all because there
are sure to be those who will circumvent it.

B) Your brief mention of dissenters is somewhat inaccurate.  I have spoken
to many people who oppose IP address transfer markets and none of them have
explained there reasoning to be that making a profit is unfair.  I think I
can fairly safely generalize most of these folks sentiment thus:  An IP
address transfer market is a radically different approach which brings with
it many new challenges and risks and it has not been shown that the
potential benefits will surely outweigh those dangers.

>
> 2) The transfers need to be registered to preserve the integrity of
> who can inject routes into the routing infrastructure - for IPv4 and IPv6.
>
> Ongoing problems with illegitimate routes being injected into the global
> routing infrastructure (either by accident or due to malicious intent) must
> be solved.  We cannot envision any way to solve this without knowing the
> current legitimate holder of address prefixes.

I fully agree that WHOIS and other registry data must be accurate.  This
starts with cleaning up the current data and then goes on to maintaining
it.  This is a needed and noble cause that I am helping to champion in the
ARIN region currently.  It has little to do with the decision to create an
IP market or not though.  If the data is accurate, illegitimate transfers
will be harder to perpetrate.  In order to hijack a route, you must
advertise it from your AS, this makes it very clear to everyone who
participates in Internet routing who is using what IPs, therefor, the
solution to the problem you raise is much more of a technical one than a
policy related one.  Take a look at the RPKI and SIDR work going on now in
the IETF for the real answer.

>
> 3) ISOC opposes a central formal managed market that clears trades and
> prices,
> for IPv4 addresses, but does not advocate barriers to transactions between
> parties.
>
> While there is potential value to operating a central clearing for
> transactions, like a stock or commodity market, especially open and
> transparent pricing, such a Market Maker would be exposed to risks of a
> variety of charges of unfairness.  It is possible (but we do not consider it
> likely) that demand for such a market will induce private market makers.
>
> What might have justified RIRs taking on those risks would be the need to
> assign addresses to fit routing hierarchy.  In such a market, prices for
> address blocks would depend on the block size, but which actual address was
> allocated would depend on the implications for the global route table as
> well as the offered price. Instead of evidence that this is necessary, what
> we have found is that the global route table is scrambled to accommodate
> traffic engineering and multihoming already, and arbitrary transfers would
> not matter.

A) While the RIRs have not done a _perfect_ job of stopping aggregation,
they have assigned addresses in a hierarchical fashion quite well.  This has
done two things; one is that they have kept routing table growth within the
ability of routing hardware over time (iow, good enough) and secondly, they
have maintained the option of aggregation.  Even though many ISPs advertise
smaller blocks for traffic engineering or other purposes (avoiding
hijacking, etc)  they  _could_  advertise much less if the need arose and  I
believe there is value in this  as we approach  IPv4 greenfield exhaustion.
B) There is a second piece to the hierarchical way RIRs have assigned and
allocated addresses, and that is the needs based component.  This is not
only to help prevent explosive de-aggregation but also to preserve openness
to new entrants.  Transfers done without the RIR supervision that we have
seen during the IPv4 allocation period have a high likely hood of allowing
large incumbents to bar entry by purchasing all available IPv4 early on and
then not releasing it to new entrants or growing but smaller competitors.

>
> 4) ISOC believes that transfers will extend the availabilty of IPv4
> addresses while
> IPv6 gets distributed.  Gradual increases in the cost of acquiring IPv4
> addresses
> may incent network operators and developers to deploy IPv6.
>
> The belief that network operators would deploy IPv6 in parallel (dual
> stack) with IPv4 while there were sufficient IPv4 addresses, so that new
> IPv6-only hosts could reach everything, was wrong.  There was no economic
> incentive for operators to prepare for a future while there were sufficient
> addresses.  It is conceivable that operators with sufficient contractual
> leverage on their suppliers and consumers could undertake the costs of
> conversion to IPv6-only in order to realize the value of IPv4 address space
> they would transfer.

By many estimates, allowing transfers would only extend IPv4 availability by
a few months if at all.  Most of the large allocations given out by the RIRs
are to ISPs.  It is not safe to assume that ISPs will sell IPv4 space at any
price because it is very likely that the buyer will be a competing ISP or a
potential customer.  Why would I sell you IPv4 space when I can rent it to
you instead?

I would like to close by stating that I am not necessarily opposed to the
current ISOC position, I just want to make certain that all the
ramifications have been taken into consideration and that the specific
agendas of any individuals involved are made clear.  The opinions I have
expressed here do not represent those of any other entity and may not even
be mine by the time you read them.
~Chris

>
> _______________________________________________
> Chapter-delegates mailing list
> Chapter-delegates at elists.isoc.org
> http://elists.isoc.org/mailman/listinfo/chapter-delegates
>

-- 
Chris Grundemann
www.chrisgrundemann.com
www.linkedin.com/in/cgrundemann
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://elists.isoc.org/mailman/private/chapter-delegates/attachments/20081024/e69be624/attachment.htm>