[arin-ppml] Pros and Cons of a liberalized transfer policy
Thomas Narten
narten at us.ibm.com
Thu Sep 4 14:36:13 EDT 2008
> Therefore, in this thread I would like some help to map out the argument
> space we are working with. I would like us to intentionally simplify the
> arguments and gloss over many of the nuances. What I'm asking for us to
> do is map out the breath and shape of the argument space we are dealing
> with here, rather than to perfectly capture the nuances of each
> argument.
Here is my summary based on what I have seen in the last 2 weeks
(and I don't think arguments have changed much in a VERY long time):
Is there anything I am missing?
To be helpful, it is probably not useful to expand on these in a lot
of detail. I think many of the frequent posters have made their
positions clear many times over. What I am interested in hearing is
whether there is another argument (pro or con) that is not covered
adequately below.
Why a revised transfer policy is needed:
- buy/selling/trading of address blocks already happens
today. Sometimes directly (recent Ebay sale attempt), sometimes
less directly (i.e., as an "asset" in a larger merger/acquisition).
To the degree that ARIN can step in and bring clarity to the market
(i.e., legitimize true ownership of a block, record transfers in
whois, in-addr.arpa, etc.), this helps/benefits the community at
large
- provide incentive to holders to return underutlized blocks. (there
is non-zero cost associated with doing this), and holders that have
space to return today have no incentive to do the work (spend the
$$) if they get nothing in return, not even cost recovery.
- we will exhaust the IPv4 free pool; transfer policies will increase
(perhaps only slightly) the availability of unused address blocks
to sites after ARIN runs out
- unclear who will sell and who will buy. Big users (large ISPs)
need large blocks, which will be in the most short supply.
- ARIN will become increasingly irrelevant if trading/selling goes on
and it has no role (e.g., with WHOIS, impacting minimum prefix
sizes, etc.). But since market pressures will cause buy/selling to
happen no matter what, ARIN will make itself irrelevent if it has
no role. When an ISP/enterprise needs space, and one way to get it
is to shell out $$, if the price is right (compared to other
options), they will do so.
Today, ARIN does have influence over block sizes and deaggration
and such. Not complete influence, but some. If transfers happen
outside of ARIN's influence, ARIN's ability to influence/curb
undesirable behavior decreases. In the worst case, ARIN has no role
whatsoever.
- The broader community not served by having whois/in-addr.arpa data
increasingly wrong. It is not the actual owner of the block that
suffers when such information is wrong, it is the rest of the
community. ARIN should be doing all that it can to maintain
accurate information.
- other RIRs will have a transfer policy doing it, and ARIN will
stick out if it does not. There is only one global internet after
all...
- IPv6 is not deployed today and won't be before exhaustion. Public
IPv4 addresses will still be needed for transition. A revised
transfer policy is part of ensuring such address space remains
available.
- Roughly 50% of the assigned space is "legacy", meaning it was
assigned before the RIRs existed. Moreover, the vast majority of
underutilized space is in the legacy area, when larger blocks were
given out to the early adopters. The vast number of legacy holders
also happen to be in the ARIN region
(http://www.nro.net/documents/presentations/jointstats.v1.0608.pdf). With
ARIN responsible for so much of the legacy space, ARIN's transfer
policy (or lack thereof) will have the most impact across all the
RIRs
- one of the key hindrances today with IPv6 deployment (or lack
thereof) is the unclear ROI. Having addressing markets - with clear
costs associated with obtaining IPv4 addresses helps externalize
the actual cost of IPv4 address exhaustion. Having a market with
real $$ amounts associated with IPv4 addresses will help justify
the cost of moving to IPv4.
- IPv4 is not going away anytime soon, regardless of IPv6
uptake. Even when/if IPv6 becomes widely used, there will still be
legacy usages. Prudent stewardship of those IPv4 addresses will
still be relevant.
- clear ownership titles for address blocks will become increasingly
important in the future. We can't secure routing, without being
able to identify who is is authorized to speak on behalf of
particular address blocks. Today, RIRs provide such information via
WHOIS and in-addr.arpa delegations. In the future, via signed
certificates. This is needed to keep to keep the internet running
securely and stably (i.e, track down spam/abuse/law
enforcement/etc.). Of course, this sort of info is only useful if
it reflects is accurate and reflects reality.
Cons:
- address trading/buying/selling doesn't happen. So, there is no
reason to do this. [yes, some people still say this.]
- nothing is needed. addresses are not property. The current system
is working just fine. (buy/selling of addresses does not happen
today) [although perhaps an extreme view, I have seen this view
expressed repeatedly]
- Liberalizing the transfer policy will cause a market to develop. We
would be facilite creation of a system that has many
downsides. Best not to go there at all.
- Legitimizing address markets will increase fragmentation of address
space, contributing to routing table bloat (which potentially
serious consequences). Note that fragementation is a result of
transfering pieces of larger address blocsk. This will happen if
trading happens on the black market. But a transfer policy may
increase the amount of transfers that take place, exacerbating the
problem.
- Not "fair" to end sites to have to pay for address space; not fair
for holders of underutilized blocks to recieve a "windfall" when
they sell addresses
related issue: roughly half of the assigned space is "legacy" and
the vast majority of underutilized space is in the legacy area.
These legacy holders reside disproportionately in developed
economies, meaning that the "windfall" would benefit developed
economies at the expense of those less able to pay (i.e, the less
developed economies, which came to the Internet after the
newcomers).
- anything that extends IPv4 lifetime delays IPv6 uptake. We should
do everything possible to encourage fast deployment of IPv6, since
that is the best overall strategy.
- the rate of address consumption is so large (dozen /8s per year)
that a market of address available via transfers is unlikely to
come anywhere near matching demand. so why bother.
- Changing the transfer policy won't really change much, so there is
no need to do anything. We will only cause ourselves grief by doing
this.
More information about the ARIN-PPML
mailing list