Jump to main contents

History of Internet Resources Management in Japan

Focusing on Domain Name and IP Address

open all section menu
close all section menu

Internet Timeline
Japanese Page

Chapter10 IPv4 address pool exhaustion and IPv6


As described in Chapter5, those involved in IP addressing and Internet infrastructure issues made a concerted effort to deploy CIDR technology and establish the IPv6 standard. This work was in preparation for IPv4 address exhaustion, which was anticipated to occur in the near future, following the commercialization of the Internet in the first half of the 1990s. Deployment of IPv6 started gradually in 1999 when IANA started IPv6 address distribution, but it stopped short of replacing IPv4, and IPv4 continued to support the rapid expansion of the Internet. Nevertheless IANA’s pool of IPv4 addresses became exhausted in 2011 and, at the time of writing, the RIRs are facing exhaustion one after another – starting from APNIC – so the deployment of IPv6 throughout the Internet is awaited with expectation. This chapter describes the effort made after the latter half of the 1990s to promote IPv6 and to counteract IPv4 address pool exhaustion in the rapid expansion of the Internet.

IPv6 emerging on the Internet

Standards for IPv6 have expanded and improved sequentially since 1995. Router vendors started to provide devices supporting IPv6, and providers attempting to construct IPv6 networks began to emerge after RFC 2460[208] (the basic specification of IPv6) was finalized in 1998. IANA started distributing IPv6 addresses to RIRs in the following year, 1999[209]. Then the RIRs, including APNIC, started allocating IPv6 addresses according to the provisional policy[210] [211].

Efforts of Japan towards IPv6 promotion

Based on the recognition that IPv6 would support the future of the Internet, the concerned parties in Japan actively promoted technological development of IPv6, both to contribute to the global Internet and to demonstrate world leadership in this field. In September 2000, the Japanese government announced the “e-Japan Strategy” in a policy speech[212] at a session of the diet made by the then Prime Minister Yoshiro Mori. This speech clearly displayed Japan’s decision to promote IPv6 in a proactive way. Also in the private sector, the IPv6 Promotion Council[213] (IPv6 Council) was formed not only by the Internet industry but also other industry segments, including the home appliance industry, which was endeavoring to equip their products with high functionality by incorporating network capability. With this, the effort toward IPv6 promotion was accelerated. Moreover, Internet Association Japan (IAjapan) also set up the IPv6 Deployment Committee[214] in April 2001 with an aim to provide a forum for promotion, education and information exchange regarding IPv6. As of writing this, the committee continues to engage in IPv6 proliferation in the Internet industry, through various activities including the IPv6 Summit.

These efforts were made not only in Japan, but also in the Asia Pacific region and at the global fora. In the Asia Pacific region, the Asia Pacific IPv6 Task Force[215] was established in 2003, and it expanded the activities to further IPv6 deployment throughout the region by enhancing collaboration and cooperation among IPv6 promotions in each country. At the global level, the IPv6 Forum[216] was established as the place for collaboration among those engaging in promotion activities all over the world. The Forum promoted IPv6 by organizing events and implementing the certification program. Takashi Arano and Hiroshi Esaki served as its board members representing Japan.

In the WIDE project, a “KAME Project[218]” for developing reference code[217] regarding the IPv6 protocol stack started, and the reference code developed through this project was used without change for operating systems based on BSD (Berkeley Software Distribution) - a UNIX operating system - and various network devices. Its implementation in other operating systems for PCs - which would be used by a large number of users - started with Windows XP SP1 and Mac OS X (10.2) that came out in 2002. Along with the KAME Project, the “TAHI Project[219]” started, and it contributed to the global verification of interoperability among the devices that supported IPv6. The TAHI Project became a certification test base for the IPv6 Ready Logo[220], one of the certification programs promoted by the IPv6 Forum.

As the proliferation of IPv6 gradually progressed, more tangible issues surfaced in terms of IPv6 address distribution that had been carried out following the provisional policy since 1999. The Japanese community, with the initiative of JPNIC IP Address Committee and IPv6 Council, also played a leading role in the review of IPv6 address policy. Discussion of the new IPv6 address policy was not directed toward developing separate policies at each RIR, since IPv6 was still at an early stage of deployment. Rather, the deliberation focused on applying a “global coordinated policy” common to all RIRs. Accordingly, it became necessary to gain agreement with each and every RIR about the policy proposal, and those who were engaged in this effort visited each region and gave presentations to attain consensus. It is said that members of the proposing group at that time traveled one and a half times around the globe for these policy discussions. The effort of the Japanese players made a significant contribution towards the new IPv6 address policy, implemented in July 2002, most of which still stands as current policy[221].

Column: Global Coordinated Policy

IP address policies are specified for each RIR, based on the consensus gained at their respective policy forum. It was around 2000 when all the three existing RIRs documented and defined this scheme. Before that, IP address management policies were specified pursuant to the standards such as RFC 1466 and RFC 2050, which were finalized through discussions at IETF. These were applied to the IP address distribution in all RIRs. After the establishment of the RIR fora, policy development was delegated to them. With regard to IPv6 addresses, the IETF separated an address into two parts: the “technical domain” consisting of the leading three bits, that is, a format prefix, as well as the last 64 bits; and the “policy domain” consisting of the remaining 61 bits. It further specified that the management policies of the “policy domain” were delegated to RIRs.

Argument for revision of the interim policy of IPv6 started just after this. Each RIR was able to determine each address policy for itself, but the conditions varied largely from region to region due to the fact that it was still at the initial stage of deployment, and the number of cases of IPv6 usage was still small. So policies for which consensus were needed could vary largely, and it was decided to revise the provisional policy, aiming for a “globally coordinated policy” under which all RIRs would adopt the same policy.

The IPv6 address policy revised in 2002 is the only globally coordinated policy as of writing this document. Since then, each RIR has made changes to their own policy, so the present IPv6 address policies do vary somewhat among RIRs. In the case of the “global policy” that defines IANA’s management procedure for IP addresses mentioned in Chapter6 and Chapter7, its definition and development process are clearly stipulated. However, the globally coordinated IPv6 policy at that time had no such stipulation, but was instead handled in an ad-hoc manner, guided primarily by the desire for a new IPv6 policy which met the needs of all the regions.

Expansion of the Internet through IPv4

Formulation of specifications for IPv6 and development of its implementation were promoted from the latter half of the 1990s to the beginning of the 2000s. At the same time, the Internet spread and evolved at an extraordinary speed. This development was primarily supported by IPv4 because IPv6 was a new protocol, and the number of users and services that were aware of IPv6 was extremely small. In addition, most services on the Internet existed on IPv4, and IPv6 was incompatible with IPv4. In this situation, it was difficult to presume that IPv6 would replace IPv4 and support the expansion of the Internet. So, while waiting for a fully-fledged development of the Internet over IPv6, distribution of IPv4 addresses was shifted to conservation mode to expand the Internet to the maximum within the limitation of available pool, which was anticipated would be exhausted in near future. Technically speaking, the classless technology described in Chapter5 and NAT (Network Address Translation)[222] technology - which uses private addresses for closed networks and converts them to global addresses for communication when connecting with the Internet - were very effective as countermeasures for the address shortage caused by the Internet expansion. Expansion of the Internet increased the need for IPv6, but it ironically raised hurdles for the migration to IPv6.

Based on these technologies, RFC 2050[223] was compiled in 1996. RFC 2050 specified that a sufficient amount of IP address space should be distributed as necessary only after a registry had closely examined the concrete IP address usage plan, in line with the conservation principle (which had been prioritized above the other four principles of address distribution, namely aggregation, uniqueness, registration, and fairness). In addition, as described in Chapter6, the principle of address conservation penetrated gradually but widely through the system in which address policies were shared among RIRs, NIRs and LIRs, and in which evaluation standards were adjusted. This is seen in mechanisms such as the “assignment window.”

Around this time, it was becoming common for companies to establish Web sites, and hosting (rental server) services became popular. The mainstream for Web sites at that time was still static content, so when assigning an IP address to a hosting service, the name-based hosting realized by HTTP/1.1[224] was recommended. For connectivity services, dynamic address assignment, under which multiple users were able to share one IP address, was the mainstream.

In this way, with assignment policy focusing on conservation of addresses, the consumption speed of IPv4 addresses slowed down slightly.

figure:allocation from IANA
Figure 10-1 The number of IPv4 addresses distributed by IANA. The consumption speed slowed down temporarily from 1995 to 2005. (as perhttp://www.potaroo.net/tools/ipv4/fig06.png)

Though stress was laid on conservation, concern over explosion[225] in the number of routes was gradually growing more serious. While the number of routes increased from 100,000 to 150,000 and on to 200,000, JPNIC changed its method of retaining its own address pool to the method of sharing the supply with APNIC. JPNIC also tried to reduce loss from division in order to avoid making addresses into small segments and to increase the address conservation effect[226].

Along with the expansion of the Internet, providers' demand for IP addresses was increasing continuously, and the overly strict evaluation criteria began to affect the business operation of providers who wanted to expand their business or start new businesses. So, distribution was to shift gradually from an exclusive focus on conservation, to considering the balance between address conservation and route aggregation.

Accelerated IPv4 address consumption through penetration of continuous connections

From the beginning of the 2000s, there was a gradual dissemination of broadband service, and the number of users expanded at a stroke from around 2004. In dial-up connections, which had been the mainstream up until then, users received an address only when they needed it. So the number of IP addresses required by ISPs to provide Internet connectivity services was smaller than the total number of users and equal to the maximum possible number of simultaneous connections. However, when constant connection became common, even with dynamic assignment, it became necessary for the number of addresses to equal the total number of users. Furthermore, to offer a service package including IP telephone and Internet connectivity, multiple IP addresses were required for one user (subscribing household) in some cases. Moreover, dynamic content or programs were run under complicated system configurations, and communications using SSL became widespread on the server side. So assignment of a unique address to each customer was required more frequently in hosting and data center services, too, and the overall consumption speed of addresses grew faster.

This happened not only in Japan but also in China and South-East Asian countries where there was sharp economic development, so the increase in demand for IP addresses became a trend in the entire Asia Pacific region.

figure:Allocation from APNIC
Figure10-2 Speed of IPv4 address distribution from APNIC. Consumption speed increased gradually from around 1997. (http://www.potaroo.net/tools/ipv4/fig27h.png)

IPv4 address pool exhaustion becomes more of a reality

Against this backdrop, forecast reports on IPv4 address exhaustion were released one after another[227] [228] in 2005, and the exhaustion of IPv4 addresses started to be considered as a real forthcoming problem among associated parties.

JPNIC closely examined the reports about IPv4 address exhaustion forecasts and validated the appropriateness of those forecasts. At the same time, JPNIC published a report entitled “Recommendations regarding IPv4 address exhaustion” in April 2006 and outlined the countermeasures against the exhaustion[229]. It became one of the triggers to have the issue of IPv4 address pool exhaustion widely recognized. However, some people, particularly those who had been using the Internet for a long time, still remained skeptical in their attitude toward the exhaustion of IP addresses itself or wondered whether it might be an excuse for forcibly promoting IPv6.

In June of the following year, 2007, JPNIC put out a press release “IPv4 address inventory depletion in Internet registries efforts”[230] and announced a position paper[231] to clarify its intention to work proactively on resolving the IPv4 address pool exhaustion issue. Subsequently, JPNIC held study meetings with executives and experts to systematically promote the review. As a consequence, the “Study report on the IPv4 address exhaustion issue (Phase I)” was published at the end of the same year[232]. This report stated that improvement of IPv4 address usage efficiency was effective only as a tentative life-extension measure and emphasized again that there would be no other alternative than accepting IPv6 as the final resolution measure.

IPv4 address pool exhaustion and IPv6 educational activity

In this way, JPNIC and the other relevant entities promoted discussions and information sharing in preparation for IPv4 address pool exhaustion. But concerned parties in Japan also considered that there was an urgent need to carry out a more expansive educational activity to raise awareness of the crisis, which could be “the largest crisis since the early days of the Internet.” So, in September 2008, various Internet related organizations, including JPNIC, cooperated to start up the Task Force on IPv4 Address Exhaustion[233]. The task force intended to make the issue of IPv4 address exhaustion known widely, considering concrete countermeasures for each stakeholder and showing this as a concrete road map to promote the handling of IPv4 address exhaustion and, therefore, acceptance of IPv6 in Japan as a whole. In some industry segments, including content providers and program developers, this information was not conveyed smoothly. However, a wealth of information was provided through activities of the task force, including an “action plan” showing the counter-acting plan for each type of organization, and issues of IPv4 address pool exhaustion were shared between participating organizations. Thus each organization supported each other through their activities and ensured that address exhaustion at IANA in February 2011[234] and in the APNIC address pool in April 2011[235] occurred without major problems or confusion.

Besides this, JPNIC provided IPv6 hands-on seminars and test beds as a core member of the initiative, to ensure penetration of IPv6 technology among network engineers.

IPv4 address policy in the face of exhaustion

With the exhaustion of IPv4 addresses becoming a reality, review of address policy started to focus on how to handle the last pool of IPv4 addresses. As a result of active discussions held at the JPNIC open policy meetings (JPOPM)[236], people who led the discussion jointly drafted the “IPv4 countdown policy”[237]. This policy stipulates that one block will be distributed to each RIR when the IANA pool is reduced to five /8 blocks. This was a proposal for “global policy” aiming at defining the operational policy of IANA. JPNIC executives and staff traveled around to promote the proposal in RIR meetings as members of the proposal team, in the same way as when IPv6 interim policy was revised. Midway through the process they got together with people of the LACNIC region who made a similar proposal, and consequently the IPv4 countdown policy was established as a global policy in 2008[238].

Immediately after the point when the exhaustion became a looming reality, the “IPv4 address transfer policy” for accommodating IPv4 addresses with other organizations by “transfer” was proposed in the Address Policy SIG[239] (Special Interest Groups) of APNIC[240]. The idea of “using IP addresses under lease from a registry” had been used consistently until then, and it was the basic idea that excess IP addresses within a certain organization would be returned to the registry for redistribution to different organizations that filed an application. On the other hand, the idea of transfer suggests that IP addresses are treated like property and allows an address assignee to determine to whom it would redistribute the addresses, which would represent a significant change in the idea of IP address management that had continued since the early days of the Internet. Therefore, significant debate arose in the APNIC forum and JPOPM. In 2009, two years after the proposal, a transfer policy for IPv4 addresses was established in both APNIC and JPNIC. Even after consensus was reached in JPOPM, JPNIC made further review at the board and put the IPv4 address transfer system into practice in August 2011[241]. With this, it became possible to transfer excess addresses of one organization to another organization, which to some extent enabled JPNIC to respond to the demand for IPv4 addresses after the exhaustion at JPNIC.

Internet over IPv6 after IPv4 exhaustion

The central pool of IPv4 addresses of IANA reached exhaustion on February 3, 2011[242]. Following this, the IPv4 address pool in the Asia Pacific region was exhausted on April 15[243]. Then, on September 14, 2012, the pool was depleted at RIPE NCC in the European region[244]. The pool of LACNIC, the regional registry for Latin American and Caribbean region, was exhausted on June 10, 2014. It was said that ARIN would face exhaustion by the end of 2014 (though at the time of writing this has not yet occurred).

In this manner, the pools at each registry are reaching exhaustion one after another, but the timing of exhaustion at the ISP and provider level varies according to the amount of addresses they hold. In addition, excess addresses are coordinated through transfer, so it is anticipated that it will take a little more time until the supply of IPv4 addresses completely disappears.

One of the planned scenarios when IPv6 was first developed was to complete IPv6 deployment before the total exhaustion of IPv4 addresses, so that exhaustion would not become a problem. In reality, while IPv6 support in the operating systems of client PCs and servers was promoted, it remained a problem for ISPs' network infrastructures. Though it would cost much to make those infrastructures compatible with IPv6, it was not easy to see the value in pursuing such an effort, thus making it too difficult to justify the expense from the business point of view. So in the specific Japanese context, the focus was on supporting IPv6 in the access lines that connect users and ISPs.

In Japan, the incumbent carrier, NTT Group, which is regulated by NTT Law, has two local service companies, NTT East and West. Together, those companies are the dominant providers of local loops for households and enterprises. NTT East and West provide “Flet’s” Service[245], an IP access service which connects users and ISPs, and most ISPs in Japan depend on it. In these circumstances, IPv6 support in Flet’s service is prerequisite for that of ISPs.

Demand for the Flet's service to support IPv6 started to appear, little by little, mainly from ISPs from around 2005, when IPv4 exhaustion was becoming a reality. Then a clear policy was announced at “Study Group on Internet's Smooth Transition to IPv6”[246] held by the Ministry of Internal Affairs and Communications from August 2007 to April 2008 and at the “Internet Policy Roundtable”[247] held from January 2008 to February 2009. During this period, discussions had been held repeatedly between ISPs and NTT, and these discussions were included in the roundtable report[248]. As a result, the method for Flet's IPv6 Internet connectivity service was approved in August 2009. The service was implemented in less than three years, and Flet’s IPv6 Internet connectivity started in May 2011, just as the IPv4 address pool of APNIC was exhausted[249].

After that, each ISP promoted support for IPv6 with their Internet connectivity using Flet's service. However, there was a problem. When a user with an Internet connection with no support for IPv6 tried to access a site that supported IPv6 as well as IPv4, very long delays occurred before data transmission started. This is called the “IPv6 to IPv4 fallback” problem.

The Flet’s service of both NTT East and NTT West had been offering a unique information service called “Flet’s Square” in addition to Internet connectivity. For this service, IPv6 was used ahead of its public use on the Internet. The fallback occurred when IPv6 closed within this Flet’s network interfered with the connection of IPv6 sites on the Internet. IPv6 was deployed on the Flet’s network in a forward-thinking manner, but ironically it resulted in trouble for developing the IPv6 Internet later on.

It was Google Inc. that led the calls for improvement of this situation, primarily because they had promoted support for IPv6 in their services since around 2008[250], but were affected significantly by fallback-related access delays.[251] Google's Japan-based staff in charge of IPv6 strongly urged NTT and Japanese providers to address and fix the problem[252].

The fundamental measure to address this issue was to ensure that users' Internet connections supported IPv6. However, it would take time for this measure to be put in place for all users. Therefore, content providers, ISPs and NTT consulted with each other and applied operational workarounds as a tentative measure.

Despite the aforementioned challenges, access lines supporting IPv6 spread widely thanks to IPv6 connectivity via Flet's, as many Japanese ISPs provided their customers with IPv6 connectivity via Flet’s service. Thus ISPs have promoted support for IPv6 in their connectivity services, while trying to solve the associated “teething problems”. As of writing, they continue to work on the development of IPv6. For instance, major ISPs provide IPv6 functionality to new users in addition to IPv4 functionality as a basic service, and some ISPs have announced a policy to add IPv6 for all existing users.

As a global effort, the Internet Society (ISOC) has taken an initiative to carry out various activities targeting providers and vendors to accelerate the movement towards IPv6. On World IPv6 Day[253] held on June 6, 2011, an attempt was made to make the main Web sites of participating organizations dual stacked with IPv6 and IPv4 for at least 24 hours. World IPv6 Launch,[254] an expanded version of World IPv6 Day, was held on June 6, 2012, and providers and vendors supporting IPv6 as a permanent operation were recruited, not just as a one-day event. Responding to the call, major content providers and ISPs including Google, Facebook and KDDI, who were already promoting support for IPv6, came forward. It appears that IPv6 traffic has been growing significantly with the aforementioned movement as a trigger, even though the proportion of IPv6 traffic to all Internet traffic is still small[255].

It is expected that IPv6 will gradually become the standard for new users in the future, and the number of IPv6 users will increase little by little. However, it will take some time and labor to ensure that the vast number of existing IPv4 Internet users and resources all support IPv6 (enabling communication with native IPv6 users). Nonetheless, this is a big turning point in the history of the Internet. Development of the Internet has been supported by a diversity of people, including not only developers and operators but also users, and this movement has overcome a number of challenges. Going forward, the history of the Internet in Japan will continue to be marked by coordination and cooperation among many people.

<< Chapter9 Ver.1.0-April 2015 Appendix1 >>

[208] Deering, S., Hinden, R., “Internet Protocol、Version 6 (IPv6) Specification”, RFC 2460, December 1995



[211] Impress R&D, supervised by Hiroshi Esaki “IPv6 text book” ISBN978-4-8443-2487-4, 8.1 Address management frontline

[212] Policy speech by Prime Minister Mori at the 150th session of the Diet

[213] IPv6 Promotion Council

[214] About IPv6 Development Committee

[215] Asia Pacific IPv6 Task Force

[216] The IPv6 Forum

[217] Standard source code implemented with technical standards

[218] The KAME Project

[219] TAHI Project

[220] IPv6 Ready Logo Program

[221] Implementation of new IPv6 policy by APNIC

[222] https://www.nic.ad.jp/ja/tech/glos-kz.html#03-nat

[223] K. Hubbard, M. Kosters, D. Conrad, D. Karrenberg, J. Postel, “INTERNET REGISTRY IP ALLOCATION GUIDELINES”, RFC 2050, November 1996

[224] Fielding, R., Gettys, J., Frystyk, H., Berners-Lee, T., “Hypertext Transfer Protocol -- HTTP/1.1", RFC 2068, January 1997

[225] https://www.nic.ad.jp/ja/tech/glos-ka.html#12-routing-explosion

[226] Assignment from APNIC address pool

[227] Geoff Huston, IPv4 - How long have we got?

[228] Tony Hain, A Pragmatic Report on IPv4 Address Space Consumption

[229] Publication of report “Recommendations regarding IPv4 address exhaustion”

[230] IPv4 address inventory depletion in Internet registries efforts

[231] Position paper “IPv4 address pool exhaustion at Internet registries”

[232] “Research report regarding IPv4 address pool exhaustion” published

[233] Task Force on IPv4 Address Exhaustion

[234] IPv4 address exhaustion at IANA and JPNIC’s address distribution for the future

[235] IPv4 address exhaustion at APNIC and address management policies to be implemented by JPNIC after exhaustion

[236] For instance, JPOPM12 held on July 17, 2007

[237] prop-046: IPv4 countdown policy proposal

[238] prop-055: Global policy for the allocation of the remaining IPv4 address space

[239] https://www.nic.ad.jp/ja/tech/glos-kz.html#03-SIG

[240] prop-050: IPv4 address transfers

[241] Implementation of JPNIC document regarding IP address management - JPNIC documents implemented along with introduction of IPv4 address transfer mechanism -

[242] Available Pool of Unallocated IPv4 Internet Addresses Now Completely Emptied

[243] APNIC IPv4 Address Pool Reaches Final /8

[244] RIPE NCC Begins to Allocate IPv4 Address Space From the Last /8

[245] Service that supplies IP network constructed by NTT East and NTT West and used for housing their subscribers. The network is provided as an underlying support line for ISPs

[246] Study Group on Internet's Smooth Transition to IPv6

[247] Internet Policy Roundtable

[248] http://www.soumu.go.jp/main_content/000009979.pdf
Page 38 (5) Coping with migration of the Internet to IPv6 (current situation 15)

[249] Access Network Working Group of Task Force on IPv4 Address Exhaustion held a reporting session at the start of the service and explained technical matters and the status of discussion.
Session materials:http://kokatsu.jp/blog/ipv4/data/wg.html
“Status and result of consideration among JAIPA, NTT East and NTT West” by Kimura, JAIPA: http://kokatsu.jp/blog/ipv4/data/02_20090615AccessWG_JAIPA.pdf

[250] Colitti, L."IPv6 transition experiences", NANOG50 Meeting, October 2010

[251] Specifically, it was pointed out that a problem would occur in the display of advertising banners.

[252] Study group on enhanced usage of the Internet with IPv6, Ministry of Internal Affairs and Communications (18th meeting), material 18-2 “Status of IPv6 in Japan and recommendations for the future (Google Inc.),” May 2012

[253] Archive: 2011 World IPv6 Day

[254] World IPv6 Launch

[255] Google's IPv6 statistical data page publishes the number of access via IPv6 to Google sites. The data shows that the access more than doubled in 2012 and 2013.