| Can a Market-Based IP Allocation
Approach Save IPv4?
By Dale Geesey
Vice President of Consulting, v6 Transition a Division of Innofone.com
 |
 |
Introduction
How long will IPv4 addresses be available? That has been a question asked
many times over the past decade and a half. In the early 1990s, projections
within the Internet Engineering Task Force (IETF) showed the imminent
depletions of the IPv4 address space under the current methods of use
and allocation. In response, a two-pronged approach was employed to mitigate
the pending crisis. The first part of the approach was to delay the complete
depletion of the IPv4 address space. This was accomplished through a combination
of techniques, including the use of more rigorous address allocation policies
and with Network Address Translation (NAT). While the delay in total IP
address depletion was successful, it was neither intended to be nor could
be the final solution. The second part of the approach was to develop
a replacement for IPv4 that would provide enough address space to support
Internet growth for the foreseeable future. The IETF initiated a working
group called IP Next Generation (IPng) that eventually led to the development
and standardization of IPv6 in the mid-1990s.
Although the delay of the IP address depletion has been successful, over
a decade later the industry is once again debating when IPv4 address depletion
will occur with estimates that the Regional Internet Registries (RIRs)
will run out of IPv4 address space between 2008 and 2013. That is unless
another method is found to once again extend the life of the IPv4 address
space. One approach that has been discussed in the past and recently reintroduced
by a major IT research and advisory firm is the idea of allowing market-based
approach to IP address allocation.
The Market-Based Approach to IP Allocation
In a February report titled, "IPv6: Unmasked" 1, Burton group
suggested that once the RIRs ran out of IPv4 address space, it would be
"necessary and desirable" to move to a market-based approach to IPv4 address
allocation. The idea being that IP address could be transferred between
user organizations. In effect, IP addresses could be treated as a commodity
that are bought and sold as necessary. Although the actual approach is
not presented in detail, the points put forward by the report by Burton
Group include:
- "Enterprises and carriers that need address space will be
able to purchase it on the open market.
- Regulatory burdens incurred when dealing with the RIRs —
namely, address space justification — will be eliminated.
- Users will be permitted to sell unneeded address space on the
open market.
- Users will no longer hoard addresses because the addresses will
be liquid assets that can be obtained easily.
- Prices will be reasonable because of the relative abundance of
address space, absent hoarding activity. (Current gray market address
space pricing is approximately $1 per address, which is likely to fall
as the supply increases in an open market.)
- The RIRs will function as registries, providing title services
— transfer and confirmation of ownership — to address holders."
The conclusion reached is that a market-based approach combined with
the continued (possibly enhanced) use of NAT and the thought that reserved
and unused space would flood into the market would extend the IPv4 address
space until 2026.
Developing the Details
As in most plans "The devil is in the details." When considering a market-based
approach many questions pop up on the specific model on how it will work.
Some of the questions include2:
- Will addresses be owned or leased?
- Will they be considered property and how will they be treated
on the balance sheet?
- Will the entire IPv4 address space be open to trading or a subset?
- What will the minimum size trading block be?
- Who will route the smaller address blocks?
- How will this impact peering table sizes?
- Will there be a centralized trading entity?
- How do we make sure there is not IP price manipulation?
Some people would argue that the general concept is the critical point
and all of the details can be worked out along the way. This is probably
true in many situations, but may not apply here. First, there is no single
governing body that can force all of the carriers, enterprises, organizations,
government and consumer users of the Internet to accept a significant
change such as this. The driving force would be for a preponderance of
carriers and governments to agree to the approach that would then drive
the remaining Internet community in willing or unwilling acceptance.
IP Hoarding: Is Less More?
Thought: A small set of organizations are hoarding a large number of IP
addresses and they would be willing to sell those IP addresses for a one-time
windfall. This is an interesting concept coupled with the belief that
the cost of IP addresses would fall once an open market is formed due
to the abundance of address space that would be available. One of the
basic flaws with this theory is that much of the existing IPv4 address
space is in the hands of governments and educational institutions that
are not driven by profit. If the US government posses 20% of the existing
IPv4 address space, then at $1 per address they could sell it all and
receive a one-time cash infusion of just over $850M. That is a nice number
and many organizations would like to have that amount of money, but it
is trivial compared to the US budget and the risk associated with not
having the address space. The Internet and its IP networks are considered
a critical asset to the US government. Consider that the estimated IT
budget for the US government will exceed $90B in the year 2010. That means
if they sold off their entire address space they would produce just less
than 1% of the IT budget for one year. Note, that this is just the IT
budget and not the trillion plus overall budget of the federal government.
A more likely scenario in this case is that other governments would take
an opportunity to purchase and hoard large amounts of IP space for their
own strategic and economic purposes. What government would not take advantage
of a limited resource that will be critical for communications and the
economic prosperity of its county to buy and hoard address space?
Carriers are another Internet constituency that would not be likely to
sell unused space, but hoard it. Anyone who has worked with customers
understands that many of them want complete solutions. There are very
large enterprises that prefer to manage their own operations, but the
majority carrier and Internet Service Providers (ISPs) customers expect
complete solutions, which will include IP addresses. IP address to a carrier
is the lifeblood of supporting customers. It will be necessary for them
to maintain a large surplus in order to support existing and new customer
demand. It would be interesting in a Market-Based IP Allocation world
if the amount of IP addresses would be tracked by analysts as one method
for valuing the future potential earnings of carriers and would ultimately
have noticeable impacts on the stack price.
CIDR: How Big or Small?
One important question in a Market-Based IP Allocation model is: How small
of a CIDR block can you sell (or buy)? The first rule of thumb is that
IP addresses will have no value unless they are routable. In a free market,
it is imaginable that anyone would be able to sell to anyone else. Thus,
the idea of hierarchical routing and limiting the routing tables is being
completely thrown out of the window. Now the question becomes; what is
the smallest number of IP addresses that can be sold that will be routed
ubiquitously across the Internet? The preponderance of business and organizations
in the US (and the world) are considered small if, as the Burton Group
suggests, the NAT model is strictly followed. Their requirement for routable
IP addresses to support general enterprise requirements would presumable
be small. In addition, they would not want to pay for address space they
do not need. A CIDR /28 (or 16 addresses) could arguably be sufficient
for the majority of small organizations using this model.
If the entire IPv4 address space is broken into separate CIDR /28s, there
are 268,435,456. Although a majority of users may only need a /28, it
is not realistic to assume everyone would, or that all of the IPv4 address
space would be broken down to this level. So if we applied a 25% reduction
factor then there would be 201,326,592 blocks of IP addresses that could
potentially need separate routing entries. This would create an enormous
explosion in the Border Gateway Protocol (BGP) tables. Although processing
power and memory are becoming cheaper, I seriously doubt if carriers could
scale up to handle the challenge without significant costs.
The IP Pricing Model
Market based pricing hinges on the Laws of supply and demand. It is a
very simple concept: Low prices occur when you have low demand relative
to supply, and high prices occur when you have high demand relative to
supply. Simply put, if there are more IP addresses available than organizations
want then the price for IP addresses will be low. If there are fewer IP
addresses available than organizations want then the price for IP addresses
will be high. One trait that would be interesting to observe would be
the use of futures on the price of IP addresses. If you have been following
the dramatic increase in oil prices, then you probably understand that
the price of oil is driven by the expected future supply and demand. That
means what you pay today for a barrel of oil (to be delivered in the future)
is based on what the market believes the conditions will be when the order
is due. How will this impact the price of IP addresses on an open market?
The interesting difference of IPv4 addresses from generally available
commodities is that it is like energy. Energy cannot be created or destroyed;
it just changes forms. IPv4 addresses cannot be created or consumed; just
change ownership. That means that there will always be a maximum potential
supply that is known and thus creates an opportunity for manipulation.
It would not be inconceivable for a group of organizations to hoard IP
addresses just to create an artificial shortage that would drive the price
up significantly. Of course this assumes that the general demand for IP
addresses are low and the market conditions would be present for a group
to reasonable come in and corner the market.
A more likely scenario is that there is a pent-up demand for IPv4 addresses
that when made generally available in an unrestricted and unregulated
basis will vie for available address space. This would then drive the
price up significantly based on natural market factors. The US, both the
commercial and government sectors, has generally enjoyed relatively easy
access to IP addresses. Although limited, they have been abundant compared
with many regions of the world. If IP addresses were globally available
how long would it be before government such as China, India or other countries
in Asia or the Middle East began purchasing them for own use or to the
benefit their country? Instead of falling, the price of IP addresses could
significantly climb over a short period of time and could become a significant
factor that could curtail growing Internet demand based on cost.
Conclusion
The basic premise behind using a free market to allocate scarce resources
for the best use based on market demand has worked exceedingly well for
the United States. It is part of the foundation of capitalism. As long
as a single or small set of people or companies cannot unduly influence
the market and the masses are able to buy, sell, and trade without significant
restriction, it can provide for a more efficient allocation and use of
available resources. But this does not mean that it is the best method
to enable an ever-growing Internet user base to continue to move forward.
There is no doubt that given a market-based approach, the companies and
government organizations that could utilize them with maximum advantage
would wind up owning the majority of the IP address space. But one of
the core concepts that had driven the Internet is a large, ubiquitous
system for communication that spans the entire world (and beyond in the
future) and that is accessible by a vast majority of the inhabitants,
not just the few who have the most money to pay for addresses.
If one approached the concept of limited address space and posed the
premise, "let's create a market where we will buy and sell,
and arbitrarily determine sets of numbers because we do not have enough,"
the logical question would be: "If we created the numbers in the
first place and there is an infinite amount of numbers, why wouldn't
we just create more?" It seems like an obvious answer and one that
has been accomplished with IPv6. Clearly there will be new costs incurred
just to set up and manage a system to support this new model and cover
the numerous legal disputes that are sure to follow. But the care and
feeding costs are not the only ones to consider. It is likely that carriers
would start charging end users directly for routing and peering their
market based IP allocation blocks to recover the costs associated with
entering and maintaining the increase in their routing tables. It is also
clear that as the cost of IPv4 addresses increased dramatically it would
drive up the cost to develop of maintain new Internet-based service offerings.
In a world where IPv4 addresses become bought and sold on the open market
there will always be addresses available, for the right price. If you
are willing to pay enough you will be able to get an IP address. The real
question is how that will impact the continued growth of the Internet
if alternative solutions, such as IPv6, are not readily available or pursued.
1Daniel Golding, "IPv6: Unmasked," Burton Group,
Feb 08, 2006
2 Daniel Golding, "IPv6: Unmasked," Burton Group, Feb 08, 2006
| 
