McFadden/Holmes Report of the Ad Hoc Group on Numbering and Addressing (8 March 2001)

The following report was prepared by Mark McFadden and Tony Holmes of the Ad Hoc Group on Numbering and Addressing. Under a 13 March 2001 resolution of the ICANN Board, it has been referred to the Address Council of the Address Supporting Organization for further consideration and development of recommendations.

February 2001
Version 1

1. Purpose and motivation

This document is put forward as a final report of the ICANN Ad Hoc Group on Numbering. It reviews the background to the formation of this group and the efforts and progress that have been made since its formation. It also includes other relevant inputs which have been updated since their initial submission to the ad hoc process. It recognises the difficulties which have been experienced as part of a factual record, in what is an area that will continue to change as the convergence between the traditional different worlds of fixed and mobile communications, data networking, and computers merge in a multimedia environment.

2. Background to the formation of the Ad Hoc Group

In the early months of 1999 proposals were being put forward regarding the creation of the Address Supporting Organisation (ASO) in accordance with ICANN's bylaws, which called for its creation as one of three advisory bodies reflecting community consensus.

One of the major goals then, and now, was to stimulate a wider public debate on IP address policies and to promote broad substantive participation in the policy making process. Accepting that IP addresses are a fundamental Internet resource without which the IP community cannot function this was clearly a key requirement. The anticipated explosion of Internet applications and network devices, the acceptance that 3rd Generation mobile networks would only serve to accelerate that demand even more, and firm plans by traditional circuit switched telecommunications networks to provision both new and existing services over IP based backbones added additional weight.

The MoU that formed the basis for ICANN to form the ASO in 1999 (see http://www.aso.icann.org/docs/aso-mou.html) was considered unacceptable by some organisations on the basis that it effectively limited the ability of a number of parties to contribute, other than from afar. Whilst it was recognised that the Regional Internet Registries 'bottom up' approach to policy formation offered one way, some parties argued that future demands, particularly those of a global nature, required a fresh approach with wider, more open participation in the ASO.

Against this background and active lobbying by a number of parties who considered they had been effectively excluded, a number of decisions were taken by ICANN. Following the formal ratification of the ASO MoU by ICANN a Resolution was passed at the ICANN meeting in Santiago to establish an Ad Hoc group, charged with the task of identifying key technology, commercial and economic drivers that would affect addressing and numbering in the Internet. This group was ask to identify issues problems, risks and opportunities, although it wasn't expected to propose, much less adopt technical solutions to them. Current trends in services, network convergence and globalisation was specifically referred to in the Resolution. The full text of the Resolution can be found at: http://www.icann.org/committees/adhoc/

3. Some difficulties experienced

The resolution also set out the working methods that were to be adopted. A small editorial team would be formed to co-ordinate a dedicated web-based ICANN public comment board. A list of those people selected by ICANN for the editorial team and the initial timetable specified, can also be found within the Resolution.

The make-up of the editorial team was carefully chosen with the intention of providing a balance from parties who held widely conflicting views at that time. Against such a background, kick-starting the work proved very difficult. An initial private exchange of views between the people involved highlighted this, with one side proposing a wide ranging study and the other claiming that most of this was 'out of scope'. This clearly underlined the fact that there could be no co-ordinated effort from the editorial team. The public comment forum was then opened, requiring those wishing to post a message to register first. The public comment forum can be found at: http://forum.icann.org/adhoc/

It soon became clear that this site was going to be used by many people as a place where they could air their grievances on a wide variety of topics, totally unrelated to the focus of the Ad Hoc Group. Things quickly degenerated to such a degree that the ICANN staff were forced to take executive action and set up an 'Off Topic Forum', which can now be found at http://forum.icann.org/offtopic/

4. Relevant inputs to the process

4.1 Input Documentation to the ad hoc web site

Two major contributions to advance the work were submitted to the public comment forum. The first was a Bibliography (submitted by Mark McFadden see: http://forum.icann.org/cgi-bin/rpgmessage.cgi?adhoc;391C7F8E0000003B). The second an input on Future Trends (submitted by Tony Holmes see: http://www.cix.org/adhoc1.html). Both documents called for comment and have now been updated. The revised versions are provided in the following sections as part of this report.

4.1.1 Current Bibliography on Addressing Trends, Standards and Drivers

Attached as appendix A is a bibliography on addressing trends, standards and drivers. The bibliography was last updated in February of 2001.

4.1.2 Future trends and driving forces - What is going on in the world?

(i) Key areas of growth and new drivers

We are now in an environment where seamless multimedia and information transport to all business and home users is becoming a part of our everyday
communications requirements. Figures indicating predicted growth appear to be updated daily and always make staggering reading. Looking at a variety of sources the following facts are put forward as a broad indication of the way things are moving.

• Traffic from the traditional telecommunications world to Internet Service Providers (ISPs) is growing at a very fast rate (in excess of 30% p.a. in many countries) and is still increasing.
• Over 5 million US consumers used the Internet for the first time in the first quarter, 2000.
• Almost 50% of the US population now have Internet access.
• Overall Internet and IP based traffic is growing at 400% a year world-wide whilst Internet revenues have been growing at more than 100% a year compared with core telecommunications businesses growth of 5% per annum.
• Data has been growing at more than 20% per year compared with 8% for voice.
• Demand for Internet access is increasing at the staggering rate of 1000% a year
• Before end of year 2000, half of all the bandwidth in the world will be Internet traffic

There is a generally accepted concept that the Internet world works on a different timescale to the rest of the telecommunications world, some say that 1 Internet year is equivalent to 10 telecommunications years. Whether this figure is valid isn't really the issue, what is true is that in terms of technological change the pace will be rapid.

Fixed Internet traffic stemming from services offered by telecommunications operators and Internet Service Providers (ISPs) will experience exponential growth. In Europe alone the fixed Internet market has been forecast to exceed more than EURO 60 billion by 2005.

Investment in cable technologies and DSL will open up a new and dynamic market to businesses both small and large, and ordinary consumers. Despite the fact that implementation is likely to vary between countries due to early teething troubles Local Loop Unbundling for instance), all predictions indicate rapid growth. Web hosting facilities and the introduction of large server farms will also fuel the rapid build out of broadband networks with huge bandwidths. According to some analysts the world-wide Web hosting market will rise from the current figure of $3 billion (E2.9 billion) to $23 billion (E22.3 billion) by 2002. Networks with a capacity measured in petabytes, (which is 1,000 terabytes) are already planned. It has been estimated that before long we may very well end up with 30Mbit/s or more per person, and the people that really need high-speed access could get bandwidths an order of magnitude higher. High initial prices for broadband will not hold back demand for long, the trend is already upwards. As an example, take-up is expected to exceed 100,00 Million in Europe by 2005.

There are a number of issues that could have an impact on the speed and direction of change (certainly in the near term), related to some specific service offerings, such as VoIP. Customers with dedicated access to the Internet have benefited accordingly, even though 'always on' doesn't necessarily mean 'always works'. TCP/IP is a best-efforts protocol and depends upon the network paths available. Providing the equivalent QoS that customers are use to on the public switched network is still an issue. Some hold the view that this can only be resolved by routeing traffic over Intranets that guarantee performance and keep unwanted traffic off their facilities. This has seen a rise in the number of carriers looking to provide service to large corporate clients who require this. Such considerations also impact back on the types of IP addresses required (public or private address space). The possible widespread deployment of RSVP (Resource Reservation Protocol) and MPLS ( Multi Protocol Label Switching) could help change this in the future.

There are also many new and potential applications that have yet to make their mark in a big way, although it can only be a matter of time. The e-commerce and entertainment sectors are already viewed as key markets, but there are also opportunities for business and industrial applications (such as remote learning and electronic publishing) telematics and infomatics and of course the home, where a variety household gadgets will also be controlled across 'the net'.

The introduction of 'Bluetooth' technology will enable digital devices such as mobile phones and desktop and notebook computers to achieve instant connectivity without using any cable connections by incorporating tiny radio receivers. Thereby facilitating transmission of both voice and data. Software controls and identity coding built into each microchip ensures that only those units pre-set by their owners can communicate. By 2002 Bluetooth functionality will be built into thousands of different devices, supporting both point to point and multipoint connections.

(ii) Voice over IP

Voice over IP or IP telephony is a particular technology for transporting and setting up voice calls over an IP infrastructure. It is now the focus of attention in the international standards arena (ETSI Project TIPHON being just one example) and is already being deployed in a number of countries. The realisation of this service owes much to technology developments, much of which was outside of the telecommunications industry. These include voice compression technology, processor power, and IP network equipment. Call control functionality will no longer be embedded within localised telephony switches, instead this functionality can be provided by network based servers which can have global significance, thanks to the adoption of internationally based standards.

The growth of IP telephony is likely to follow the trend of Internet access growth as the incremental to provide this application on a PC is minimal. Breakout from/to the PSTN will also occur (either near- end or far-end) so for the foreseeable future both IP addresses and E.164 numbers (traditional PSTN/ISDN) will also be required for voice telephony. Ultimately current economics indicate voice telephony will migrate from traditional PSTN networks towards international broadband multi-service (IP based) networks. Given that multiple service providers will be supporting IP based telephony, it is essential that current efforts to set in place address resolution mechanisms are completed. Carrying voice traffic over corporate IP based networks means that only one network is required. Voice over IP will continue to grow as ADSL based 'always on' connectivity is rolled out.

(iii) The Mobile sector

We are already experiencing huge growth in the mobile sector.

The current market size is shown in the figure below.

(Source UMTS Forum 2000 report)

It is estimated that in excess of 78% of Internet users also use mobile communications and that this trend will only increases with 3rd Generation Mobile handsets that will bring IP to the handset.

The predicted growth of the mobile market is shown below.

 

 

When you consider that in some areas of the world current penetration is currently very low, such as less than 5% in the Asia Pacific region, then even these predictions could be surpassed. The ratio of mobile to fixed is changing rapidly and they are no longer viewed as a supplement to the fixed networks. Many developing countries are building mobile networks as a less expensive means of providing access, for example Cambodia. As mobile operators amortise the build out cost of their new networks and new licensees generate increase competition mobile call prices are expected to fall. As voice has been the prime service to date this will be particularly noticeable in that market sector.

Much of the focus now is on 3rd Generation Mobile (or UMTS). Mobile multimedia combines real-time voice, image and textual information.

Mobile growth and Internet access capability

Source: UMTS Forum

At the time this report was compiled the latest survey predicted that 535 million cellular telephones will be sold in 2001, an increase of 33 per cent over sales in 2000.

Bluetooth technology will also offer new ways of using mobile devices both for professional and personal use. As well as being able to access e-mail, web browsing, video conferencing and a range of other varied services will be possible. Users will come to expect access from their mobile terminals to a similar range of multimedia applications that can be reached from traditional fixed network connections. There will also be additional benefits for the mobile user such as access to time critical information related to stockmarket prices and on line share dealing, traffic information, etc. Work is already underway to provide tailored applications so that consumers will not feel they are trying to browse the web whilst looking through a keyhole because of the smaller screens on mobile devices. Applications will remain a key enabler but with consumer applications likely to grow very quickly with the introduction of WAP (Wireless Application Protocol), coupled with existing demands for telemetry applications such as meter readings, alarm and stock control, this area will advance rapidly.

The provision of 'always-on' capabilities for users will add yet another dimension along with the development of the new services and e-commerce applications.

Merill Lynch estimates that growth in mobile data is expected to be 70% p.a. in the next 5 years. Data traffic will eclipse voice traffic on wireless networks by 2004. There is also a view that multimedia data will account for up to 60% of total mobile traffic in 2005 with more people accessing the Internet from mobile devices as opposed to desktop by that date.

What is clear is that UMTS will facilitate the integration of traditional mobile communications capabilities with data information capabilities and access to the Internet and Intranets which will fuel demand. Initiatives like 'Bluetooth' and the rapid introduction of WAP technology, coupled with new types of terminals and digital assistants will serve to increase the demands on the world of IP naming and addressing.

GPRS

This trend has already started with the introduction of GPRS (General Packet Radio Service). This is an enhancement to the existing digital GSM voice based network. It conveys data across the mobile network using IP packet based switching. With its 'Always on' functionality its considered to provide a more efficient use of network resources. It provides higher data rates, up to 40kb/s initially although this should more than double with time, but even the initial offering is far higher than the existing GSM circuit switched technology which provides 9.6 kb/s. GPRS is a stepping stone towards high bandwidth 3rd Generation mobile networks (3G/UMTS). With its focus on data networking and new opportunities for e-commerce on the move, GPRS has also had to build in a range of access control and data security measures. This is yet another area that will demand constant attention as the new mobile market grows.

The first GPRS Network was launched by BT Cellnet in June 2000 enabling corporate users to access corporate intranets. Deutsche Telekom's mobile operator T-Mobil later announced they would be providing GPRS services in Germany by September 2000.

GPRS will evolve through several phases Initially GPRS will provide a core network overlay of IP to the circuit switched phase 2 GSM. This can facilitates IP access from mobile terminals to an ISP or corporate network using tunnels. The radio interface used ensures compatibility between mobile networks whether they are GPRS compliant or not. With the introduction of UMTS Release 5 (3 phases later) we will see the core migrate to a single IP network and the addition of multimedia services based on IP. Initial releases allow a choice of IP version 4 or IP version 6 in the PS (packet switched) domain, although UMTS Release 5 (forecast implementation 2003/4), will specify that IPv6 is used for the IM (IP multimedia) domain. The IM domain can be considered as a standalone area of the network. Currently all GPRS networks use IPv4. Basically the three factors that could drive UMTS migration towards IPv6 are exactly the same as those for other types of networks; additional benefits from v6 capabilities, v4 address exhaustion or equipment standards.

3rd Generation Mobile

3G Mobile embraces a wide family of networks, including those that are referred to as UMTS and IMT2000. As well as providing voice these networks will facilitate high speed data access, up to 20 times faster than current technology. On demand bandwidth and wide multimedia functionality. Video conferencing, video imaging, video on demand and downloading information using 3G terminals will be everyday occurrences. The greatest growth is expected to be in the areas of content provision, terminal manufacture and service provisioning. Areas that were traditionally the focus of mainly ISP operators, such as those related to Internet naming and addressing, will increasingly become a major focus for mobile network operators and those concerned with service provisioning. The number of players in this market place is expected to mushroom over the next few years.

Auctions for 3G licences will remain a focal point over the next few years as administrations identify their preferred way of handling a heavy potential demand from both existing and new entrants in to the mobile market. Whilst Finland awarded licences in a 'beauty contest' the UK chose to use the auction process, raising in excess of $35 Billion in the process. Licence costs are now proving unpredictable in spectrum auctions across Western Europe with prices ranging between 100 EURs and 150 EURs per capita in the early rounds. Whether regulatory authorities choose to allocate through auctions, beauty contests or lottery it will not stifle the race to open up this new market sector, neither will the cost. If companies are forced to pay large sums of money to obtain licences they will need to achieve payback as soon as possible. This could well lead to innovative marketing and widespread arrangements with 3rd party service providers and other mobile networks.

Implementation of 3rd Generation Mobile seems certain to occur first in Western Europe, North America and Asia. Understandably the most developed countries will lead the way. Europeans being particularly aggressive with the push towards licence assignment. The importance with which this is viewed is underlined by the fact that virtually all existing operators have competed vigorously, and it most cases successfully, for a licence in order to safeguard their ambitions for the future. In the US it was originally thought that existing PCS networks would evolve towards 3rd Generation technology, but this has seen them start to fall behind the pace of change occurring elsewhere.

The table below provides some indication of the world-wide focus on 3G licensing.

Country	Licence procedure defined	Timetable established for bids	Licences awarded
Australia	X	X	X
Austria	X	X	X
Belgium	X	X
Canada	X	X
Denmark	X	X
Finland	X	X	X
France	X	X
Germany	X	X	X
Greece	X	X
Holland	X	X
Hong Kong	X	X
Ireland	X	X
Italy	X	X	X
Japan	X	X	X
Luxembourg	X	X
Netherlands	X	X	X
New Zealand	X	X	X
Norway	X	X
Portugal	X	X
Spain	X	X	X
Sweden	X	X	X
Switzerland	X	X
UK	X	X	X

The introduction of new players in the guise of Mobile Virtual Network Operators (MVNOs) should also increase growth in this sector. This concept enables new players to enter the market without the burden of having to obtain licences or spectrum or the additional costs this incurs. New interconnect arrangements will enable them to negotiate for a variety of resources which will then result in them bringing to the market place a variety of services, without providing much of the infrastructure required. They can adopt a business model and operate in manner which can be sustained by the maturity of the market they choose to operate in. They may take different roles, such as service integrators, bundling airtime, Internet access and a variety of different applications. Its generally considered that this will fuel the market and generate growth in an area where start-up costs could otherwise jeopardise benefits gained through competition.

The IMT2000 terrestrial and radio interface specifications will further optimise performance in a wide range of operating environments using satellites within the LEO, MEO and GEO orbits.

(iv) e-Commerce

E-Commerce is set to revolutionise the way we do business. Its facilitator again being the new communications technology. All aspects of trading will be influenced, from the ordering stage, through the supply chain and of course even the actual transfer of money between involved parties. As technology becomes more pervasive organisations will adapt and collaborate within a fast emerging global e-commerce market. Existing Internet sites such as Yahoo and Excite are merely early examples. Whilst specialist communications tools such as EDI (Electronic Data Interchange) has facilitated e-Commerce for many years this has mainly provided links between large companies with established trading relationships. The rapid introduction of highly intelligent devices at low cost now brings this capability within reach of all consumers with the Internet being the key as the ubiquitous common trading mechanism. Entry costs are low and no one on the web is more than a click away. Some research organisations estimate that business transactions over the Internet will be in excess of $300bn this year. World-wide acceptance of non cash trading methods such as credit cards coupled with a readily available global, always open, market pace (the web) will also fuel e-Commerce.

Currently growth figures for on-line purchasing vary between regions, but much of this is dictated by PC ownership and the take up of Internet access. On-line sales also vary greatly between products, fast moving and convenient consumer goods (travel, books, CDs, stocks) being at the top end. Compared to the US. Europe and Asia still have a long way to go before credit card penetration is comparable, but the pace of change is already increasing. Recent surveys have indicated that e-commerce revenues for European SMEs will soar by 800 percent, from just over EUR400 million last year to over EUR3 billion by 2003. Much of this due to on-line selling.

Concerns over security also have to be overcome Consumers must be confident that deals are not visible to third parties and that authentication between communicating parties is secure. Rapid advance in encryption techniques and the use of cryptograph and asymmetric key systems should do much to elevate such fears. Public key schemes provide confidentiality by encrypting documents so that they can only be read by the owner of a corresponding key, whilst digital signatures can facilitate authentication and transaction integrity. Major industry players including Mastercard and Visa are also working to develop SET (Secure Electronic Transactions) which should provide a secure method of paying by credit card over the Internet.

The introduction of smart cards will provide a convenient way of providing key storage and the generation of digital signatures, as opposed to storage on hard disks. Smart cards have the potential to bring together a wide range of applications in e-Commerce. Future generations enabling financial service companies to offer a single card combining traditional credit and debit card facilities alongside strong authentication and an electronic purse capability. It has been estimated that up to seven billion smart cards will be in circulation by 2004.

As well as the technical developments taking place the other significant barriers to growth which need to be tackled are the legal and regulatory aspects. Currently the world of e-commerce is still getting to grips with the demands stemming from the administrative and legislative sectors. New frameworks are still the focus of much attention by International trade organisations and other external bodies such as the European Commission. Industry codes of practice are also likely to play an ever increasing role. Already a number of governments have announced their intention to introduce legislation making digital signatures legally binding.

The addition of speech recognition software in to palmtops and mobile handsets will also assist in growing the market place. Prototypes already exist and commercially available products will follow shortly. Intelligent content management will also be very important, to enable consumers to navigate through a huge amount of information and choice.

Against this background the power of the Internet has been widely recognised. Information fuels free markets. Consumers have already identified the ability to exploit price disparities between countries, which will ultimately result in price levelling, bringing clear benefits. The inevitable tide of e-commerce will mean that today's dominant player who fails to keep pace will be bypassed. With existing geographic constraints being removed it is essential that the resultant growth of the Internet, which will be spectacular, is always considered on a global basis.

(v) Global Networks, Intranets and VPNs

Within the telecommunications environment a number of international carriers and ISPs are looking to serve the need of large multi-national companies. Global alliances and strategic partnerships are also bringing a new dimension to the market place as a rapid build out of large global networks takes place to meet future telecommunications needs in a ubiquitous manner. The provision of sophisticated value add data services being key in order to meet the demands of the e-commerce driven market. Much of the required bandwidth will now be provided over IP based infrastructure. Out sourcing arrangements will also require these global companies to meet new and emerging requirements in a consistent way, no matter what part of the globe they are operating in. Managed global IP based VPN's will become the norm, providing guaranteed bandwidth and performance levels. Intranets will greatly enhance business communications, both internally and with key business partners, providing a much more effective way of gaining and distributing information.

(vi) Home Networking

Home networking is also viewed as a future key growth area. Entertainment and information appliances are rapidly increasing in the home. 'Networked' devices will increase in places such as the kitchen, the garage and in other areas for security and control functions. We will also see the introduction of devices as diverse as refrigerators and TVs becoming part of the network through residential gateways. Many devices will share resources between them and also with the outside world, some via high speed data connections. Today the main appliances tend to be PC's, set top boxes, games machines and modems The interfaces are many, but a few tend to be more ubiquitous such as UBD (Universal Serial Bus) Ethernet and possibly IEEE1394. Virtually all new consumer PC's and peripherals are equipped with a USB interface, and many leading brands are providing IEEE1394 interfaces. Ethernet has been around for a long time and many business ADSL and set top boxes have these interfaces built in.

(vii) ENUM

Within the telecommunications world much attention is currently focusing on an initiative which will enable E.164 numbers (public telephone numbers) to be mapped into the Internet's Domain Name System (DNS). Through this mechanism, telephone numbers can be employed by users to look up a variety of addresses, including those used for phone, fax and email, by which the target user can be contacted. This also enables end users to tailor the manner in which they be contacted through a single E.164 number. This additional contact information can also be easily added to or amended without changing the number used for access. It will enable the provision of a variety of new and emerging applications to be accessed from circuit switched networks through the use of normal telephone numbers.

'ENUM' is the acronym that has been adopted to describe this capability, it is also the name given to the Internet Engineering Task Force (IETF) Working Group that defined the protocol.

This initiative will facilitate both-way interworking between the PSTN and Internet Protocol (IP) based networks; e.g., by using ENUM to translate a number into a domain name and then using this to look up a corresponding IP address to complete the connection. It therefore provides a method of interconnection between the traditional circuit switched world (PSTN) and the new world of IP based networks which could help to increase the pace of change and the move towards using IP infrastructure for traditional telephony focused applications.

(viii) TIPHON

The European Telecommunications Standards Institute, ETSI established TIPHON (Telecommunications and Internet Protocol Harmonization Over Networks), to support the market for voice communication and related voiceband communication between users. It also aims to ensure that users connected to IP based networks can communicate with users in Switched Circuit Networks such as PSTN, ISDN and GSM based networks, and vice versa Given the universal nature of IP networks, the prime goal is to produce global standards. It is working on schemes to permit network operators to implement guaranteed Quality of Service levels, and new billing models. Work is also carried out to define security, call set up and signalling standards relate to both the H.323 protocol and SIP. Networks based on TIPHON developed specifications are due to launched in 2001.

5 Some early global addressing requirements

5.1 GPRS Request for IP addresses

At RIPE 35 (Amsterdam, 22 - 25 February 2000) a representative of BT Cellnet (on behalf of the GSM Association gave the first of his presentations on "IP Addressing for GPRS Data Network Infrastructure and International Networks" There was considerable debate regarding the suitability of the use of and the amount registered addresses required, and the potential use of RFC 1918 addresses. A number of issues and concerns were raised and it was evident that a better understanding both the infrastructure and requirements was required. A small task force was therefore formed. Their main objective was gain agreement on an approach that could be used by all GSM operators to use public registered addressing in the GPRS infrastructure. Thereby enabling GPRS roaming services to be supported. This met in April 2000 and the report of this meeting can be found at http://www.ripe.net/ripe/wg/lir/gprs/index.html. The approach agreed at this meeting was then submitted to the LIR working group at the RIPE meeting in May 2000 as part of the process for its approval and steering it through to adoption. At this meeting it was acknowledged that some operators were already in the process of configuring their GPRS networks to support a commercial service by Q2 2000, for which globally unique addressing was required. In order to meet this demand it was accepted that initially any public address space already assigned to operators for use in the GPRS infrastructure could be utilised.

At RIPE 36 (Budapest, 16 - 19 May 2000) a follow-up presentation was made. There was agreement in principle to the use of IPv4 registered address space to identify nodes in the GPRS network infrastructure. In summary it has been agreed that

• Public IPv4 address space can be used in parts of the GPRS network infrastructure.
• Existing IP address allocation policies and procedures apply.
• Requests from mobile network operators can be sent directly to the RIPE NCC or their data network backbone providers

Address requirements for mobile devices (handsets) are still not been finalised and work continues on determining the needs of GPRS terminals and third generation mobile systems. The results of the GPRS Task Force can be found at: http://www.gsmworld.com/cgi/bounce/www.ripe.net/ripe/wg/lir/gprs/

Having gained agreement with RIPE there was still a need to consult with the other RIRs, ARIN and APNIC.

For APNIC this happened at the APNIC Meeting held in Brisbane, Australia (October 2000) Consensus agreement was achieved in principal.

The issue was raised at ARIN at their meeting in Herndon Virginia (October 2000). No final decision was taken on whether the use of public address space was valid, the issue being referred to a mailing list for further discussion. Minutes at http://www.arin.net/minutes/public/arinvi/ARIN_VI_PPM.htm

5.2 Addressing requirements for UMTS

In December 2000 the UMTS Forum, an association of telecommunications operators, manufacturers and regulatory authorities, as well as IT industries and media industries world-wide, produced a report 'Naming, Addressing & Identification Issues for UMTS'. This initial study looked at three types of addresses, IP, Mobile station roaming numbers and routeing prefixes for E.164. In this document only the IP addressing issues are considered. The report stated IPv4 address will be needed for the core network infrastructure because it is the default protocol for UMTS Releases 3 & 4. It was also recognised that a new version of the mobile operators internal DNS, compatible with IPv4 and IP V6 (e.g. BIND9.X) will be required when IPv6 is used in the core networks. The report proposed that the GSM Association should facilitate the introduction of IPv6 by providing mechanisms for operators to exchange information about v6 introduction and the transition tools they will use.

The report set out the need for IP addresses to be used for end points within the GPRS/UMTS network infrastructure, for mobile terminals connected across the mobile networks to ISPs and mobile terminals in multimedia services. In the absence of any other reason to start earlier, the report states that UMTS operators will have to start to introduce IPV6 in the IP Multimedia domain when they start to implement services based on Release 5 of the UMTS standards, around 2003-2004. The PS domain for UMTS Release 5 can be either IPv4, IPv6 or dual stack.

6. IPV4 - IPv6 Evolution

IPv6 will be most apt for supporting new mass-market IP services because of its carrier scale capabilities. The move from IPv4 to IPv6 is most likely to occur as an evolutionary path, which is likely to pick up pace as the amount of available IPv4 address space reduces. Such a transition is not a trivial exercise. A number of studies have been conducted trying to predict when IPv4 address exhaustion will occur. Various methods have been utilised, but there is no single 'correct' method of predicting such an event when there are so many variables to be taken into account. The continued growth of the Internet and IP address usage such as that set out in Section 4 of this document makes any assessment a moving target. On that basis predictions based on two separate approaches are put forward for consideration.

6.1 How much IPv4 address space is left?

The starting point for any assessment has to be an assessment of how much address space is still available. Data used in this report may be found at the IANA Website at http://www.isi.edu/in-notes/iana/assignments/ipv4-address-space

The total IPv4 address range is 000/8 through to 255/8. The current list of non-usable Internet address space (special use prefixes) given below, is taken from "Documenting Special Use IPv4 Address Blocks that have been registered with IANA" - by Bill Manning.

0.0.0.0/8

10.0.0.0/8

127.0.0.0/8

169.254.0.0/16

172.16.0.0/12

192.0.2/24

192.168.0.0/16

all D/E space i.e.224/8 through to 239/8.

The remaining address space is assumed to be usable.

6.2 IPv4 Address Space currently allocated to the RIRs

• RIPE has 7 x /8's (62, 193, 194, 195, 212, 213, 217)
• ARIN has13 x /8's ( 63, 64, 65, 66, 199, 200, 204, 205, 206, 207, 208, 209, 216)
• APNIC has 6 x /8's (61, 202, 203, 210, 211, 218)

IANA still has many address blocks reserved and available for allocation to RIRs, namely:

1, 2, 5, 7, 23, 27, 31, 36, 37, 39, 41, 42, 49, 50, 58, 59, 60, 67-95, 96-126, 197, 201, 219-223, i.e. 84 /8s. In addition to this, the former "Class B" space is completely free from 173/8 to 191/8, i.e. 19 /8s. This makes a total of 103 /8s available for allocation to the RIRs.

6.3 Assessment method 1

Method 1 focuses on IP address demand based on predictions for Cellular IP devices plus DSL lines and cable modem - it excludes NAT deployment. For best case prediction it factors in extensive use of WAP proxy servers and DHCP, thus reducing the demand on registered IP address space. Most likely case scenario factors in what is considered to be the most likely realistic deployment of proxy servers and DHCP, whereas worst case factors in very limited but realistic deployment of proxy servers and DHCP. DHCP gain from dynamic allocation of addresses is factored in as 1:1 business user and 5:1 residential user. It is important to note that both "NAT" and "no-NAT" solutions are available to the ISPs, the extent to which each option is deployed will vary from ISP to ISP on a global basis. It must be appreciated that these figures reflect one just one set of opinions and should be therefore be treated as subjective.

The H ratio not used in these calculations. Instead of this an 80% utilisation has been built in to the original figures i.e. the original figures have been scaled up by 25% to give the values shown in the Tables in §4.1, §4.2, §4.3 and §4.4 below.

Reclamation of allocated but unassigned IPv4 address space is a possible option, but one that is considered to be quite problematic and is thus not considered in this report, although it would clearly extend the lifetime of IPv4 address space.

6.3.1 Global: Cellular IP (millions) - address demand is based on terminal demand

1999/2000	2000/2001	2001/2002	2002/2003	2003/2004	2004/2005
284	398	520	640	875	1,250	worst case
144	200	260	319	438	625	most likely
29	40	53	65	88	125	best case

Sources of Data

• various sources including GSM Association which is based on EMC World Cellular Database at http://www.gsmworld.com/membership/ass_sub_fore.html
• BT Market (and other) Intelligence Reports and the appropriate forecasts for GPRS and UMTS.
  

6.3.2 Global: DSL+ Cable Modems (millions) - address demand is based on line demand

1999/2000	2000/2001	2001/2002	2002/2003	2003/2004	2004/2005
22.75	61.38	127.5	186.3	230.25	285.6	worst case
12.5	31.25	63.75	93.75	115.0	143.8	most likely
2.25	6.123	12.75	18.625	23.0	28.5	best case

Sources of Data - various sources including BT Market Intelligence Reports, and other forecasts for DSL + cable modems. Industry analysts are consistently increasing significantly their estimates of DSL deployment, as have Carriers (figures will probably need to be revised upwards).

6.3.3 Global Address Demand for Cellular IP Plus DSL + cable modem (millions)

1999/2000	2000/2001	2001/2002	2002/2003	2003/2004	2004/2005
306.75	459.4	647.5	826.3	1,105	1,536	worst case
156.5	231.25	324.0	412.75	553	769	most likely
31.2	5 46.1	65.75	83.625	111	154	best case

This table simply represents the sum of the two previous tables

6.3.4 Global Address Demand for Cellular IP plus DSL + cable modem ( /8s)

1999/2000	2000/2001	2001/2002	2002/2003	2003/2004	2004/2005
18.28	27.38	38.59	49.25	65.86	91.5	worst case
9.33	13.7	19.31	24.6	32.96	45.84	most likely
1.86	2.75	3.92	4.98	6.62	9.2	best case

This table represents the previous table in units of /8 s - 80% utilisation is built in to these figures

6.4 Method 1 - summary of exhaustion dates

The following dates are illustrated in Fig. 1 below;

Best case (late exhaustion) 2014 (for 84 x /8s) and 2017 (for 103 x /8s)

Most likely case Q3 2006 (for 84 x /8s) and Q1 2007 (for 103 x /8s)

Worst case (early exhaustion) mid 2005 (for 84 x /8s) and mid 2006 (for 103 x /8s)

As a benchmark the current world population is 6 x 109. IPv4 addressing range is 4 x 109 in comparison to the IPv6 addressing range is (4 x 109) x (4 x 109) x (4 x 109) x (4 x 109).

Fig 1 - Global demand for Cellular and ADSL/Cable modem - exhaustion dates (best case, most likely, and worst case)

6.5 Assessment method 2

Method 2 is a completely different approach that focuses on the current rate of global allocation of IPv4 registered address space, and makes clearly stated assumptions about predicted global rates of allocation. It is independent of whether private addressing is used and does not factor in NATs, proxy servers or DHCP. This method simply considers the rate of consumption of the remaining "/8" address blocks, with no assumptions regarding which market sectors are creating the demand or consuming the address space. This method is based on the current rate of allocation of IPv4 Registered address space globally from the three RIRs. It considers rates of 1.3, 1.5 and 1.8. global projections on various rates of allocation of IPv4 address space from RIRs to LIRs. Future work on this method will be based on historical data and compound growth rate over previous years rather than just the % increase on the previous year. The historical data is currently being gathered.

3 assumed rates of growth (global)
Series 1 - factor of 1.3 increase per annum year after year
Series 2 - factor of 1.5 increase per annum year after year
Series 3 - factor of 1.8 increase per annum year after year
Note
Year ending Dec 99 - global rate of increase was 1.8
Year ending Dec 00 - global rate of increase was 1.3

This method makes no assumptions regarding which services or applications are creating the demand and consuming the address space. This method does not factor in NATs, proxy servers, or DHCP - it simply varies the rate of global allocation.

6.5.1 Series 1 - Global factor of increase 1.3 per annum (year after year)

• Global rate of consumption of 4.68 /8s per annum at end of 2000
• Two scenarios (i) 84 x /8s remaining and (ii) 103 x /8s remaining

12 month time frame	factor on previous year	/8s consumed in this year	running total of /8s consumed "to date"
end 00	1.3	4.68	4.68
end 01	1.3	6.08	10.76
end 02	1.3	7.91	18.67
end 03	1.3	10.28	28.96
end 04	1.3	13.37	42.33
end 05	1.3	17.38	58.70
end 06	1.3	22.59	81.29
end 07	1.3	29.37	110.67

6.5.2 Series 2 - Global factor of increase 1.5 per annum (year after year)

• Global rate of consumption of 4.68 /8s per annum at end of 2000
• Two scenarios (i) 84 x /8s remaining and (ii) 103 x /8s remaining

12 month time frame	factor on previous year	/8s consumed in this year	running total of /8s consumed "to date"
end 00	1.3	4.68	4.68
end 01	1.5	7.02	11.70
end 02	1.5	10.53	22.23
end 03	1.5	15.79	38.03
end 04	1.5	23.69	61.72
end 05	1.5	35.54	97.26
end 06	1.5	53.31	150.29

6.5.3 Series 3 - Global factor of increase 1.8 per annum (year after year)

• Global rate of consumption of 4.68 /8s per annum at end of 2000
• Two scenarios (i) 84 x /8s remaining, and (ii) 103 x /8s remaining

12 month time frame	factor on previous year	/8s consumed in this year	running total of /8s consumed "to date"
end 00	1.3	4.68	4.68
end 01	1.8	8.42	13.10
end 02	1.8	15.16	28.27
end 03	1.8	27.29	55.56
end 04	1.8	49.13	104.69
end 05	1.8	88.43	193.12

6.6 Method 2 - summary of exhaustion dates

The following dates are illustrated in Fig. 2 below;

Increased based on increased rate of 1.3 per annum year on year; 2006 (for 84 x /8s) and 2007 (for 103 x /8s)

Increased based on increased rate of 1.5 per annum year on year; mid 2005 (for 84 x /8s) and end 2005 (for 103 x /8s)

Increased based on increased rate of 1.8 per annum year on year; mid 2004 (for 84 x /8s) and end 2004 (for 103 x /8s)

Fig 2 - Global demand based on assessed rates of global allocation

6.7 Comparison of projected exhaustion dates for each method.

Method 1	Method 2
Best case (late exhaustion)· 2014 ( for 84 x /8s ) 2017 ( for 103 x /8s )	Increase of 1.3 per annum year on year end 2006 ( for 84 x /8s ) mid 2007 ( for 103 x /8s )
Most likely case Q3 2007 ( for 84 x /8s ) end 2007 ( for 103 x /8s )	Increase of 1.5 per annum year on year mid 2005 ( for 84 x /8s ) end 2005 ( for 103 x /8s )
Worst case (early exhaust) Q4 2005 ( for 84 x /8s ) Q1 2006 ( for 103 x /8s )	Increase of 1.8 per annum year on year mid 2004 ( for 84 x /8s ) end 2004 ( for 103 x /8s )

6.8 Conclusions

Two very different methods have been reviewed, the first based on projections for terminals and lines (not addresses), the second based on projected registered IP address allocations. Whilst its accepted that method 1 best case is very difficult to extrapolate, in general there is close agreement between both sets of results in what is considered the mid-range case and the "worst case' scenarios for both approaches. Its also accepted that as yet there is little industry consensus on this issue, or a killer application that alone will drive a transition to v6 in the short term, but what cannot be denied is that demand for address space will continue to grow. All major vendors have stable Beta code available and equipment specifications are looking to include v6 capabilities. Against such a background the figures put forward above do not appear outlandish.

What is critical is that the industry works together in a co-ordinated manner to address the outstanding issues and to set in place policy and procedures that meets the needs of the fast emerging global market place.

7. The ASO and its policies

The Address Supporting Organisation (ASO), is one of three ICANN Supporting Organisations. The purpose of the ASO being 'to review and develop recommendations on Internet policy and structure in relation to the system of IP addresses, and to advise the ICANN Board on these matters'.

An archive of the ASO meetings can be found at: http://aso.icann.org/meetings/

At the November 2000 ICANN meeting an update on the ASO (Address Council) was presented which not only covered the role of the AC and its make up but also included issues for the wider addressing community. This recognised that whilst the role of the AC was not to develop policy it should in fact oversee and co-ordinate the policy development process facilitated by the Regional Internet Registries. This will become even more important once LACNIC and AfriNIC are fully operational. It was also shown that work was underway by each or the RIRs to review the other RIRs policy and procedure documents and to identify differences as well as looking to address questions over the openness of the AC. The presentation can be found at: http://aso.icann.org/meetings/other/LA2000/index.html

This presentation also identified a number of issues that needed to be considered by the addressing community. On that list were global v regional address policies, rapid market changes like GPRS, questions over the emerging RIRs, IPv6 deployment and the need to understand IPv4 address consumption. Some of those issues have been covered to varying degrees within this report.

The fact that market differences can lead to different policy needs in different parts of the world was also put forward. This is not disputed, but what does need to be fully understood is 'why and where'. If a policy is one market is likely to restrict development and growth in another market, it is certainly isn't viable to promote a single global approach. Accepting this as the 'norm' without any justification is equally difficult.

The presentation concluded with an acknowledgement of the activities realised in the past year under the umbrella of the ICANN Ad Hoc group on addressing, and it was particularly noteworthy that a specific reference was made to the members of the Ad Hoc editorial group to take part in the regional policy for a discussions concerning these issues.

Its also worth noting that in their presentation to the GAC (Government Advisory Committee) at the ICANN meeting in California (November 2000), when talking about global policy formation, the three existing RIRs stressed the need for open policy forums with 'final approval by the community'. It acknowledged the need for responsive policy development that was 'fair to all' and that co-ordination by RIRs on global policy was a prime requirement for globalisation of the Internet, accepting that local implementations may differ must policies must converge.

[8]. The way forward

In line with the decision made at the ICANN meeting at Marina del Rey, California (13 to 16 November 2000) the Ad Hoc group will be put forward for formal closure at the ICANN meeting in Melbourne, Australia (9 to 13 March 2001), following submission of this closure report.

Since the formation of the Ad Hoc group the methodologies, working practices, procedures and relationships between the RIRs, the Address Council and the Address Supporting Organisation have become clearer. The work of the Ad Hoc group has produced a detailed bibliography pointing to documents that describe the intersection between addressing requirements for IP and other technologies as well as undertaking a wide review of future drivers and address space requirements. The Address Council has acknowledged the need for this work as well as the need for additional policy co-ordination and more involvement of the wider addressing community.

When coupled with the acceptance that an open and transparent approach is a fundamental requirement, particularly where global address policy formation is the goal, this effort provides a clear indication that close co-operation and involvement of all sectors of the industry is a necessity. With the emergence of additional Regional Internet Registries this will become even more important.

Having completed the task set for the Ad Hoc Group it is proposed that ICANN should continue to seek assurance, and regularly test, that the goals set for the Address Supporting Organisation when it was originally conceived, are fully met within the existing arrangement.

Click here for Exhibit A