Section 7.2 APNIC

Gerard Ross

2013.2.17

The Asia Pacific Network Information Centre (APNIC) – Formation and early operations (1992-1995) - DRAFT

For the past two decades, the Asia Pacific Networking Information Centre (APNIC) has served as the Asia Pacific’s regional Internet registry (RIR), “charged with ensuring the fair distribution and responsible management of IP addresses and related resources ... [which] are required for the stable and reliable operation of the global Internet”.^[1] During that period, APNIC has also become an increasingly significant hub of Internet community activity in the region and a respected voice of the Asia Pacific Internet community on the global stage.

Understanding how APNIC formed, the role it serves, and the way it operates requires an overview of the architectural, operational, and administrative aspects of the Internet and the general principles which underpin all facets of Internet development.

Early evolution of IP addressing

At its most basic, the role of the Internet is to move data packets from a source to a destination. To deliver a packet, the Internet needs to know where the destination is (the address) and the best way to get there (the route). Internet Protocol (IP) addresses support these needs by identifying both the network and host.

The original Internet addresses were 32 bits long, with the first 8 bits of the field used for the network part of the address, leaving 24 bits for local address. Although more than 4 billion addresses were possible, the fixed 8 bit network part restricted the possible number of networks to only 256.

By the late 1970s, even though the total address space for the “internet experiment” was abundant, it was clear that the time had come to “prepare for the day when there are more than 256 networks participating in the internet”.^[2]

So, in 1981, the Internet Engineering Task Force (IETF) modified the address architecture to allow for three classes of Internet address. “Class A” addresses allowed for 128 networks with 24 bits of local addressing; “Class B” addresses allowed 16,384 networks with 16 bits of local addressing; and “Class C” allowed 2,097,152 networks with 8 bits of local addressing.^[3]

The new Internet Protocol version 4 (IPv4) immediately eased the limitation on new networks joining the Internet, but its “classful” addressing architecture (also referred to as “subnetting”) sowed the seeds for two new, interlinked problems that emerged with the growing popularity of the Internet.

Using classful addressing, if a network operator needed to address 200 hosts, then a single Class C network would neatly solve the operator’s need. However, if an operator needed to address 2000 hosts, then one could assign either a single Class B and waste more than 63 000 addresses, or assign eight separate Class C addresses. While the latter solution would not waste addresses, it would add eight new entries to the global routing table.

In the early days of the Internet, these issues were not critical, but as the Internet developed and commercial interest loomed, the tension between the need to conserve addresses and the need to aggregate routing information led to the next step in the evolution of the addressing architecture, namely the move from subnetting to “supernetting”.

Supernetting became a standard in RFC 1519^[4] as “Classless Inter-Domain Routing (CIDR)”; however, the issue had first been explored in detail in RFC 1338, published in June 1992, which warned:

As the Internet has evolved and grown ... in recent years, it has become painfully evident that it is soon to face several serious scaling problems. These include:

1. Exhaustion of the class-B network address space....

2. Growth of routing tables in Internet routers beyond the ability of current software (and people) to effectively manage.

3. Eventual exhaustion of the 32-bit IP address space.^[5]

RFC 1338 proposed doing away with classful addressing and instead relying on “variable-length subnetting” to allow any size of network allocation. Under this scheme, a variable-length subnet mask (more commonly known as the “prefix”) replaces the fixed-length network part of the address. Therefore CIDR allocations can be tailored to closely fit network operators’ address needs without waste, while adding only a single entry to the global routing table.

CIDR proved to be a timely and effective solution to what was by then one of the Internet’s most serious challenges,^[6] extending the life of the IPv4 address pool by two decades while allowing the routing system to scale at a manageable rate. However, it is important to realise that CIDR’s success was not solely due to architectural developments, but also relied heavily on responsible administrative practices and inclusive, community-based policy making.

Expansion of the registration function

The Internet depends absolutely on the integrity of its addressing system. But the addressing system can only work if IP addresses are unique, and uniqueness can only be guaranteed by an effective registration system.

In the early days of the Internet, address registration was a straightforward function. Network operators – of which there were few, and all of which were (generally) known to each other – simply asked for addresses, the details of which were recorded by the Internet Assigned Numbers Authority (IANA).

The IANA function was, for many years, a manual task performed by one man – the late Jon Postel.^[7] Over time, as this task grew, the IANA responsibility for “numeric network and autonomous system identifiers” was formalised and “a single Internet Registry (IR) was designated: the Defense Data Net Network Information Center (DDN-NIC) at SRI International”.^[8]

By 1990, the IETF’s Network Working Group noted:

With the rapid escalation of the number of networks in the Internet and its concurrent internationalization, it is timely to consider further delegation of assignment and registration authority on an international basis. It is also essential to take into consideration that such identifiers, particularly network identifiers of class A and B type, will become an increasingly scarce commodity whose allocation must be handled with thoughtful care.^[9]

RFC 1174 proposed that the IANA and the IR functions would remain centralised, but that “the IR would also allocate to organizations approved by the Coordinating Committee for Intercontinental Research Networking (CCIRN) blocks of network and autonomous system numbers, as needed, and delegate to them further assignment authority”.^[10]

General Internet growth and associated increases in demand were not the only factors that put pressure on the centralised registration function. The nature of a CIDR-based addressing scheme required far greater administrative oversight of address allocations. Because allocation sizes could be of any size, to fulfill the address conservation goal, it became necessary to more carefully analyse the requestors’ needs. And because variable-length subnet masking allowed multiple contiguous ranges to be announced as a single route, allocation practices had to allow for networks to grow while fulfilling the address aggregation role.^[11] Furthermore, as the complexity of the request and registration process increased, so did the need for regionalised service with greater multilingual support.

RIPE NCC emerges

Meanwhile, in Europe, Internet Service Providers (ISPs) had formed Réseaux IP Européens (RIPE) as “a collaborative organisation ... to ensure the necessary administrative and technical coordination to allow the operation of a pan-European IP network”.^[12] The first RIPE meeting was held in Amsterdam in May 1989.

RIPE is a community and has no legal structure. However at the fourth RIPE meeting in March 1990, Daniel Karrenberg presented on the community’s administrative needs. The minutes note that:

A discussion followed on the what model should be used, either a centralised or a distributed framework. What became clear from this was the need for some sort of formalised body for performing the functions of a NIC.^[13]

By September that year, this discussion had become a formal proposal to create the RIPE Network Coordination Centre (NCC) to support the RIPE community and “function as a ‘Delegated Registry’ for IP numbers in Europe, as anticipated and defined in RFC 1174”.^[14] The RIPE NCC formally commenced operation in April 1992.^[15]

The APNIC experiment

The Internet’s distributed development model has always been based on inclusive collaboration and open sharing. Individuals and organisations have always come together and cooperated to assume the various operational and administrative tasks required to develop and maintain core Internet functions.

One such forum for cooperation – more significant in the Internet’s early days than now – is CCIRN, established to provide “a forum for members to agree and progress a set of activities to achieve inter-operable networking services between participating international entities to support open research and scholarly pursuit.”^[16]

On 10 June 1992, CCIRN held a meeting in Tokyo, Japan,^[17] which is particularly notable as a confluence of several emerging factors. This was CCIRN’s first meeting in Asia, and those present included:

• Kilnam Chon^[18] of Korea Advanced Institute of Science and Technology (KAIST), a leading pioneer of Internet development in Asia

• Jun Murai of Keio University, founder of the Widely Integrated Distributed Environment (WIDE) project, and commonly referred to as the “father of Japan’s Internet”,^[19] and

• several representatives of the RIPE community and RIPE NCC, which had officially launched only two months earlier

The minutes of the meeting note Chon’s report that APCCIRN (the Asia Pacific grouping of CCIRN, then being formed under his leadership) had held two preliminary meetings but now needed to hold its first official meeting as soon as possible to start coordinating networking activity in the region. Chon also reported that there were at that time “10‐12 countries with network connection” in the region and “two international networks; PACCOM and CAREN”.^[20]

The meeting heard reports from the RIPE community, including a full report of the fledgling RIPE NCC by its manager, Daniel Karrenberg. Steve Goldstein of the US National Science Foundation noted that RIPE NCC’s work was outstanding and set an example for Network Information Centre (NIC) operation “for anywhere in the world”.^[21]

Jun Murai also reported on progress at the Japan NIC (JNIC^[22]): He explained that WIDE, Todai International Science Network (TISN), and others had established JNIC for domestic IP address and .jp namespace assignments, but that it “would be possible to expand the role to cover Asian Pacific, if desired”.^[23]

From this meeting, the elements were in place for the birth of APNIC.

Seven months later, in January 1993, APCCIRN held its first official meeting in Honolulu, Hawaii, with the Asia Pacific Engineering and Planning Group (APEPG). At this meeting, attendees discussed and accepted a plan for a “NIC for APCCIRN Region (APNIC)” to carry out resource registration, information provision, NOC support, and NIC cooperation. The minutes note that it “was decided for JNIC (M. Hirabaru and J. Murai) to carry on APNIC experiment with cooperation of other countries including Australia, Korea and New Zealand”.^[24]

The “APNIC experiment” earned a mention in the notes of the March 1993 IETF User Services Area meeting (along with the emergence of the new InterNIC).^[25]

Then at the second APCCIRN meeting – held on 8 September 1993, immediately following the INET93 meeting in San Francisco – APNIC was discussed in detail. Murai presented an overview of the global and regional NIC situation and noted that the “vision now is for an global NIC with regional authority delegated to regional NICs, such as InterNIC in the Americas and RIPE NCC in Europe”. KRNIC and JPNIC each presented status reports of their domestic projects. Murai proposed changes to the original proposal and explained:

If approved by the APCCIRN, this pilot project would help determine how to meet the needs in the region over the long-term. During the pilot phase, prime focus would be on the Internet Registry and Routing Registry functions. Limited attention would be given to informational functions until after a decision on a long-term approach for the region. JPNIC has agreed to provide resources for the pilot project.^[26]

The updated proposal^[27] was adopted and the APNIC pilot project was chartered to begin operation on 1 September 1993 and end on June 30, 1994. The APNIC pilot project goals were:

• determine the requirements for a regional NIC and the means to meet those requirements

• implement a regional IP address allocation strategy in accordance with RFC-1466

• provide a testbed for experimentation into network coordination in the Asia Pacific region

• coordinate with local, national, and regional NICs

• experiment with tools used to support NIC operations^[28]

The approved document “Asia Pacific Network Information Center Pilot Project Proposal” became one of the first official APNIC documents (APNIC-003). In addition to the goals listed above, this document also notes that the APNIC pilot project would be entirely funded by JPNIC, which had committed 10 per cent of its own budget for the work. APNIC-003 also makes it clear that the pilot project was to be “coordinated under the auspices of the APCCIRN” and that the project staff would provide a mid-term and final report to APCCIRN. At this stage, APNIC’s proposed staffing requirements were for a quarter-time manager, a half-time technical support role, and a full time administrative support role. The individuals who had expressed interest in working on the project were:

• Masaya Nakayama, University of Tokyo

• Masaki Hirabaru, Kyushu University

• Taeha Park, KAIST

• David Conrad, Internet Initiative Japan.^[29]

In the late 1980s and early 1990s, David “Randy” Conrad was a young researcher working at the University of Hawaii, Manoa on the NASA/NSF-funded Pacific Communications (PACCOM) project, which interconnected the networks of universities and research institutions in Japan, Korea, Hong Kong, Australia, New Zealand, and the US.

PACCOM held an annual meeting in Hawaii where the various researchers would meet to discuss Internet issues. It was at those meetings that Conrad met some the Internet pioneers from the region, such as Jun Murai, Geoff Huston, John Houlker, Robin Erskine, and Kilnam Chon. From this contact, Murai invited Conrad to move to Japan to help launch the Internet Initiative Japan (IIJ). In 1991, Conrad accepted the invitation, moved to Japan, and became IIJ’s seventh employee. However, once there, Conrad found that IIJ was facing some delays getting the licences it needed to provide Internet services. “I had spare time,” Conrad explains, “and Jun Murai asked if I'd be interested in helping to start up APNIC”.^[30]

With Conrad as manager, the APNIC pilot project started operations on 1 September 1993. An email from Masaya Nakayama the following day contains details of the first APNIC mailing lists:

• apnic-all@apnic.net (for all those interested in the APNIC pilot project)

• apnic-member@apnic.net (for actual APNIC pilot project members)

• apnic-staff@apnic.net (for the pilot project staff)^[31]

In APNIC’s early operations, most of its work was based on and around those lists (hosted on a computer at the University of Tokyo). In an article from 1994, Conrad explained how the work was coordinated:

The staff mailing list implementing most of the APNIC functions currently consists of 25 people from the countries of Australia, China, Japan, Korea, New Zealand, Singapore, Taiwan, Thailand, and the US. This informal group comes to a rough consensus on requests for information and/or address space assignment and fulfills those requests typically within one to two working days.^[32]

A JPNIC newsletter article from late 1994 gives more detail of the breakdown of responsibilities, explaining that the “APNIC WG” (working group) – coordinated out of JPNIC and comprising representatives of each national NIC – played the major role in allocating IP addresses. The Korean NIC (KRNIC) was in charge of “information services”. And the Australian NIC (AUNIC) maintained APNIC’s DNS-related services.^[33]

This informal “staff” arrangement was motivated by the “immense size of the Asia and Pacific Rim regions, from the Persian gulf area to the island nations of the South Pacific, and the vast diversity of cultures, religions, and economic situations encompassed in this space,” and so “one of the basic assumptions held by the members of the APNIC pilot project staff ... was that the ultimate APNIC would need to be highly distributed”.^[34]

On the subject of infrastructure, the pilot project equipment requirements were modest, with the proposal calling for “a single workstation class computer with sufficient disk and network capacity to fulfill the needs of the project”.^[35] JPNIC provided this machine in the form of a Sun Sparcstation, described as “fully connected to the Internet”.^[36] The APNIC computer served mailing lists, FTP and gopher archives, the Asia Pacific network database and whois server (which originally ran the InterNIC’s rwhois server software), and a DNS server providing referrals to APNIC services, including:

• in-addr.arpa (reverse delegation) services hosted at AUNIC, and

• an “experimental” web site hosted by KRNIC.

Again, as Conrad explained at the time, by “distributing the NIC services, the APNIC pilot project has been able to take advantage of talents of personnel at the national NICs and thus does not require extensive expertise to be located centrally at an APNIC facility”.^[37]

The dominant thinking at this stage in APNIC’s formation was that its future was as a functionally distributed organisation,^[38] with a small central coordinating centre ensuring consistency and continuity. The model envisaged was that APNIC’s various operational functions would be contracted to national NICs for fixed periods and rotated through the region.^[39] Indeed, an article by Conrad noted that APNIC, “when it becomes permanently established in July of 1994, actually will consist of a group of semi-independent cooperative organizations”.^[40]

Overtly, this approach was seen as a way of building technical capacity across the region, while also providing a sustainable operational model in the absence of an alternative established funding model. However, this approach could also be seen as a way of defusing the potential political tensions that may arise from permanently settling such a critical function in one part of the region.

The appropriate scope of APNIC’s functions was also an active area of discussion during the pilot project. On 10 December 1993, the APNIC staff published a proposal for APNIC to assume the following roles:

1. Allocate Class C network blocks to national registries and multi-national service providers

2. Provide an initial contact point for people interested in internetworking in the Asia Pacific region

3. Promote Asia Pacific internetworking

4. Provide NIC services for nations with no regional NIC

5. Promote the establishment of national NICs

6. Provide a top level information repository

7. Delegate the 202, 203, and 204.in-addr.arpa reverse domain name trees

8. Provide a forum for coordination between regional NICs in the Asia Pacific region^[41]

Of these proposed roles, the first and fourth are worth particular attention here, because they suggest a function that is significantly and specifically different to the way APNIC’s primary role eventually emerged.

Until the early 1990s, much of the Internet activity around the world had involved some form of national coordination by a NIC (either formal or informal in nature). This of course had a natural fit with the domain name system, which features country code top level domains (ccTLDs). However, as classless, CIDR-based IP addressing was introduced, national boundaries lost much of their significance and address administration practices relied on network topology rather than physical geography. The topological approach to addressing remained for the rest of IPv4’s life span and continues as an essential aspect of IPv6 addressing.^[42]

Nevertheless, even by the end of the pilot project, APNIC retained a strong focus on promoting the formation of national NICs to join the ranks of Australia (AUNIC), Japan (JPNIC), Korea (KRNIC), and Taiwan (TWNIC).^[43] National NICs were seen as a way to mitigate the “limits on the APNIC's manpower, language, and budget resources”, and a proliferation of NICs across the region were considered “better able to serve the end

users in their particular countries, and also relieve the APNIC of some operational burden by distributing the workload”. The final pilot project pilot report stated that “this promotion activity will remain one of the APNIC's prime objectives”.^[44]

Although the operational relationship between APNIC and national NICs eventually emerged in quite a different form, these considerations can be seen as the root of the crucial, yet sometimes complex, role of National Internet Registries (NIRs) in the Asia Pacific region.^[45]

All other priorities notwithstanding, the central thing a regional Internet registry (RIR) needs to justify its existence is a pool of addresses. And so, on 1 April 1994, the IANA publicly recognised APNIC’s status by delegating the IPv4 address ranges 202/8 and 203/8.^[46]

The end of the experiment – lessons learned

The APNIC experiment formally ended on 30 June 1994. Shigeki Goto (NTT and JPNIC) reported to the CCIRN meeting in Amsterdam that month that APNIC was serving 27 members from 12 countries and had assumed authoritative delegation for the 202 and 203 Class B address blocks. Goto also reported that “unresolved issues include guidelines for establishing national NICs, further service delegations and especially funding”.^[47]

Before IANA delegated address ranges to APNIC, the project typically received up to five enquiries per week. After the IANA delegation, that jumped to “5 to 10 email messages and 1 to 5 fax messages a day”, and it was clear that the work rate would grow quickly as the Internet expanded in the region and that running a NIC permanently would require “significant investments in both time and talent”.^[48]

Furthermore, as Conrad explained:

Since allocations of network addresses must be carefully considered, with special emphasis placed on ... allocating the appropriate amount of address space in a way that conforms to the requirements of CIDR, the APNIC project staff has had to explain Internet addressing and the global routing table situation many times, sometimes more than once to the same individual. In the AP region, this sort of situation can have added complexity due to language and cultural differences and typically must be handled with some care.^[49]

In his characteristically laconic style, Conrad concluded that “running a regional NIC is not for the weak of heart, shallow of mind, or shy of disposition”.^[50]

The more formally-worded final report of the APNIC pilot project was prepared by Vince Gebes of the AT&T Jens Spin Project in Tokyo, covering all major aspects of the background and operational status of the project. However, the three most significant strategic areas of concern Gebes identified were the needs to:

• define the Asia Pacific region and ensure better coordination of responsibilities between APNIC, RIPE NCC, and the InterNIC

• create guidelines for establishing new national NICs

• finalise funding models and future operations (which Gebes identified as “clearly the most important issue” facing APNIC)^[51]

These issues notwithstanding, the APNIC pilot project was considered a success and from 1 July 1994, the work continued as the “interim APNIC” until a more permanent model could be developed.

Community engagement and the challenge of sustainability

As was the case with all major Internet forums and institutions, the birth of APNIC was an expression of community collaboration, with stakeholders from many fields and nationalities contributing time, energy, resources, and talent. Building and sustaining such a community requires more than just mailing list activity, so, in addition to the ongoing registration and informational services, the interim APNIC began planning for its first major public event.

In late 1994, APNIC announced that the first APNIC meeting would be held at Chulalongkorn University in Bangkok, Thailand on 16 and 17 January, 1995. The announcement noted that “Service Providers, national NICs, and other interested individuals are invited to attend and participate”.^[52] The invitation note is significant: while all RIRs have now developed fee-paying membership structures, their meetings about addressing issues have always been open to all interested participants. Openness and transparency are core values of the broader Internet development process and are cornerstones of Internet address policy.

Furthermore, the first official APNIC address request form encouraged all network operators to get involved in the technical community and the address policy making process^[53], a function for which APNIC would subsequently become the regional forum.

The APNIC 1 meeting was well attended, with 60 registered participants. Some of the more notable attendees include:

• Masaki Hirabaru (JPNIC)

• Sanjaya (PT IndoInternet, Indonesia)

• Robin Erskine (Australian National University)

• Che-Hoo Cheng (Computer Services Center, The Chinese University of Hong Kong)

• Dr Tan Tin Wee (Technet Unit, Computer Centre, National University of Singapore)

• Tommi Chen (Technet Unit, Computer Centre, National University of Singapore)

• Shigeki Goto (JPNIC and NTT)

• Pindar Wong (Hong Kong Supernet)

• Masaya Nakayama (JPNIC)

• Xing Li (Network Center, China Education and Research Network)

• Taeha Park (I.Net Technologies and KRNIC)

• Barry Raveendran Greene (Singapore Telecom and SingNet)

All of these individuals made significant contributions to APNIC’s formation and establishment. Several have also continued long and productive relationships with APNIC and, in some cases, have served on the Executive Council or assumed leadership roles on the APNIC staff.

At APNIC 1, Conrad reported on the status of APNIC. At that time, APNIC had one part-time staff member (though part-time in this case meant at least 40 hours per week), supported by 44 others in 15 countries on the apnic-staff mailing list. The organisation used office space provided by Keio University and IIJ and computer hardware on loan from IIJ, NTT, and WIDE. In its brief operational history, APNIC had processed 311 address requests, representing 1082 networks.^[54]

However, one of the major issues discussed at this meeting was just how APNIC should continue, in particular, how it should be funded. Conrad explained in frank terms that the APNIC operational model to date had the advantage of being inexpensive, but suffered from the associated “very poor quality control” and “highly variable services”.^[55]

Conrad reviewed the main options – a donation model (which was unpredictable) versus a fee-based model (charging for either addresses, services, or time) – and compared the models used in RIPE NCC and the InterNIC. Until then, although APNIC had encouraged donations from ISPs depending on size (small US$1500, medium US$5000, and large US$10000), no ISPs had made such a donation.^[56]

Apart from inherent instability, the donation-based funding model highlighted the importance of APNIC’s perceived independence. For this reason, APNIC made it clear that it would “refuse to accept donations from an organization which [expected] preferential treatment”.^[57]

Another fundamental stumbling block towards a sustainable model was APNIC’s legal status, which had so far been informal and unincorporated.^[58]

The meeting reached a general consensus to develop a more stable funding model. RIPE meeting minutes from later that month report Conrad explaining that “APNIC raised the issue of charging for address space or leasing out address space to applicants, at a rate of approx. USD 0.10 per host address.... If it were to be implemented it would generate pressures on other registries to do something similar.”^[59]

At the same RIPE meeting, Conrad noted APNIC’s proposed use of ISO-3166 country codes to aid in defining the region. This decision has persisted and the ISO-3166 designations have become the standard for country identification used by all current RIRs.

In February 1995, APNIC published a formal overview of its organisational structure and operations. The document explains that during the pilot project phase, APNIC derived its authority from APCCIRN (by then renamed to APNG^[60]). However, because APNG was an “informal group with no legal authority or claim to represent the entire AP region” it could not provide the interim APNIC with an appropriate legal umbrella. “As such, it was felt by the members of the APNIC consultative committee that APNIC's authority should be derived from the Internet Assigned Numbers Authority (IANA).”^[61] This arrangement provided sufficient legal protection, but did not solve the funding problem.

Meanwhile, in response to the growing workload, APNIC took on its first new employee, Yoshiko Chong Fong, a Japanese national and graduate of Computer Science from the University of New Mexico. Chong was hired just before the second APNIC meeting, held in Honolulu with the APNG meeting on 2 July 1995^[62]. With little practical experience of Internet addressing, Chong received intensive on-the-job training, starting with the “APNIC meeting in Hawaii, followed by four weeks at InterNIC, one week at an IETF meeting in Stockholm, another two weeks at InterNIC, and four weeks at RIPE NCC in Amsterdam. This had prepared me to become the first hostmaster of APNIC”.^[63]

Throughout 1995, alongside the normal operational activities, a great of APNIC’s work revolved around efforts to define and implement a new organisational arrangement and funding model. In September, that work culminated in the release of a new funding plan, based largely on the RIPE NCC model. The plan included an account startup fee of US$1,000 and several payment tiers:

• Internet Service Provider Registries:

- Large – US$10,000 per year

- Medium – US$5,000 per year

- Small – US$2,500 per year

• Enterprise Registries – US$1,500 per year

Registries would be free to select their own payment tier, and the plan stated clearly that APNIC was charging for service levels, not for resources.^[64]

The funding plan was originally intended to be implemented on 1 September 1995. However, because the issue of APNIC’s lack of legal status remained outstanding, implementation was delayed until “after the third APNIC meeting to be held in Singapore in January, 1996” by which time APNIC would have established “a clear and well understood legal umbrella”. Until then, the Internet Society (ISOC) had agreed to act as a legal proxy able to receive funds on APNIC’s behalf.^[65]

By the end of 1996, APNIC’s operations – if not organisational model – were solidly established. The staff, comprised of David Conrad (half time), Masaya Nakayama (half time) and Yoshiko Chong Fong (full time), had made 747 allocations for a total of 45,473 networks. APNIC forecast a 1996 budget of approximately US$247,000, with most increases from the 1995 budget flagged for more staff resources to handle system administration, research and development, and administrative functions.^[66]

APNIC’s other major achievement in 1995 was to lead the creation of the Asia Pacific Regional Internet Conference on Operational Technologies (APRICOT), which has now become the major Internet operators’ forum for the region. According to the APRICOT website, “David Conrad ... was the instigator and prime motivator to make APRICOT the Internet Technical and Operations meeting for the Asia & Pacific region”.^[67] From 1996 on, the first APNIC meeting of each year has been held with APRICOT, bringing together the best of the region’s operational and address policy community members.

Though much remained to happen before APNIC evolved into its current form – and moved to its current location in Brisbane, Australia – the achievements of its first few years were significant. Many individuals made valuable contributions, many of which were voluntary^[68]. However, in the words of Kyoko Day (who in 1996 became APNIC’s next full time employee):

There is one thing I know for sure. Without David Conrad, APNIC did not have a ground base to start. He was only 28 when he took over a pilot project. He worked day and night. His devotion and commitment made a big difference. He took time to listen to the voices of the key people in the community but maintained APNIC core values: neutrality and impartiality.^[69]

^[1] “About APNIC”, APNIC website, February 2013. http://www.apnic.net/about-APNIC/organization

^[2] Clark, D., and Cohen, D., “A Proposal for Addressing and Routing in the Internet,” IEN 46, June 1978. http://www.ietf.org/rfc/ien/ien46.txt

^[3] Postel, J., “Assigned Numbers,” RFC 790, September 1981. http://tools.ietf.org/html/rfc790

^[4] Fuller, V., Li, T., Yu, J., and Varadhan, K., “Classless Inter-Domain Routing (CIDR): an Address Assignment and Aggregation Strategy,” RFC 1519, September 1993. http://tools.ietf.org/html/rfc1519

^[5] Fuller, V., Li, T., Yu, J., and Varadhan, K., “Supernetting: An Address Assignment and Aggregation Strategy,” RFC 1338, June 1992. http://tools.ietf.org/html/rfc1338

As an aside, RFC 1338 proposed solutions to only the first two of the three problems identified. The third problem, that of IPv4 exhaustion, was addressed much later with the specification of IPv6.

^[6] The argument could well be made that most of the Internet’s most serious problems are symptoms of its unprecedented popularity and success.

^[7] “John Postel”, Wikipedia, February 2013. http://en.wikipedia.org/wiki/Jon_Postel

^[8] Cerf, V., “IAB Recommended Policy on Distributing Internet Identifier Assignment and IAB Recommended Policy Change to Internet 'Connected' Status,” RFC 1174, August 1990. https://tools.ietf.org/html/rfc1174

^[9] ibid.

^[10] ibid.

^[11] Karrenberg, D., Ross, G., Wilson, P., Nobile, L., “Development of the Regional Internet Registry System”, The Internet Protocol Journal, Vol. 4(4), December 2001. http://www.cisco.com/web/about/ac123/ac147/archived_issues/ipj_4-4/regional_internet_registries.html

Note, this article – written by various RIR staff members – contains a more detailed overview of the RIR system in general; the establishment of APNIC, ARIN, and the RIPE NCC; and the emergence of AfriNIC and LACNIC.

^[12] RIPE NCC Quarterly Report, Issue 2, September 1992. http://www.ripe.net/internet-coordination/press-centre/publications/annual-reports/ripe-ncc-quarterly-report-issue-2

^[13] Minutes of RIPE 4, 22-23 March 1990. http://www.ripe.net/ripe/meetings/ripe-meetings/ripe-4

^[14] Blokzijl, R., Devillers, Y., Karrenberg, D., Volk, R., “RIPE Network Coordination Center (RIPE NCC)”, RIPE-019, September 1990. ftp://ftp.ripe.net/ripe/docs/ripe-019.txt

^[15] Karrenberg, D., Ross, G., Wilson, P., Nobile, L., (2001)

^[16] CCIRN website, February 2013. http://ccirn.net

^[17] Minutes of the CCIRN meeting, 10 June 1992, Tokyo, Japan. http://ccirn.net/general/TokyoJune1992.pdf

^[18] In 2012, the Internet Society announced Professor Kilnam Chon as an inaugural inductee in the Internet Hall Fame as a “Global Connector”. http://www.internethalloffame.org/inductees/kilnam-chon

^[19] “Jun Murai”, Wikipedia, February 2013. http://en.wikipedia.org/wiki/Jun_Murai

^[20] Minutes of the CCIRN meeting, 10 June 1992.

^[21] ibid.

^[22] Now JPNIC. See http://www.nic.ad.jp

^[23] Minutes of the CCIRN meeting, 10 June 1992.

^[24] Minutes of APCCIRN/APEPG meeting, 12-13 January 1993, Waikiki, Honolulu. https://sites.google.com/site/internethistoryasia/lib/add-doc/apccirn-document-list/apccirn-021

See also this APCCIRN “areas of work” document for a list of the individuals tasked to work on the APNIC experiment: https://sites.google.com/site/koreainternethistory/lib/arc/anc/anc-93-086

^[25] Meeting notes March 1993 IETF User Services Area meeting. ftp://ietf.org/ietf/93mar/area.userservices.93mar.txt

^[26] The official minutes of this meeting could not be located. The exact nature of the source document used for this information is not clear, but was provided by Professor Masaya Nakayama of Japan and appears to be a report of the APNIC discussion at the APCCIRN Honolulu meeting. http://ring.u-toyama.ac.jp/pub/doc/jpnic/minutes/committee/199309/930917-shiryou2-6.txt

^[27] ibid.

^[28] Conrad, D., “The Asia Pacific Network Information Center, Present and Future”, Connect, 1994. http://ftp.apnic.net/apnic/info/articles/connect

^[29] Murai, J., Conrad, D., “Asia Pacific Network Information Center Pilot Project Proposal (APNIC-003), 1 September 1993. http://ftp.apnic.net/apnic/archive/apnic-003-v001.txt

^[30] Conrad, D., Personal correspondence, 8 November 2012.

^[31] Nakayama, M., archived email, 2 September 1993. https://sites.google.com/site/internethistoryasia/lib/apccirn-apng/wg-bof-and-mailing-lists/apnic-all

^[32] Conrad, D., “The Asia Pacific Network Information Center, Present and Future”, 1994, Connect. http://ftp.apnic.net/apnic/info/articles/connect

^[33] JPNIC report on the APNIC pilot project, March,1994. http://www.nic.ad.jp/en/pub/unix-m/1994-03.html

^[34] Conrad, D., “The Asia Pacific Network Information Center, Present and Future”, 1994.

^[35] Murai, J., Conrad, D., “Asia Pacific Network Information Center Pilot Project Proposal” (APNIC-003), 1 September 1993.

^[36] Asia Pacific Network Information Center Pilot Project Midterm Status Report (APNIC-006), 10 December 1993. http://ftp.apnic.net/apnic/archive/apnic-006-v001.txt

^[37] Conrad, D., “The Asia Pacific Network Information Center, Present and Future”, 1994.

^[38] “Proposal for Distributed APNIC Operations” (APNIC-007), 8 December 1993. http://ftp.apnic.net/apnic/archive/apnic-007-v001.txt

^[39] ibid.

^[40] Conrad, D., “A Nic For The Asia-Pacific Region”, 22 April 1994, NSF Network News, republished in CINET-L Newsletter Issue No. 17, Saturday, April 30, 1994. http://www.nsrc.org/db/lookup/report.php?id=890202396097:497433169&fromISO=CN

^[41] “Various Proposals for the Asia Pacific Network Information Center (APNIC-004)”, 10 December 1993. http://ftp.apnic.net/apnic/archive/apnic-004-v001.txt

This, the KRNIC proposal, and the Mid-Term report were presented and discussed at the APCCIRN meeting in Taiwan in December, 1993.

^[42] Unfortunately, the misunderstanding of the difference between national and topological addressing strategies persists and remains at the heart of many Internet governance disputes today.

^[43] Notably absent from this list is China, which in early 1994 still had very little TCP/IP networking in place, though momentum was rapidly building for its emergence as a major contributor to the Asia Pacific Internet community.

^[44] “Asia Pacific Network Information Center Pilot Project Final Report (APNIC-009)”, 1 July 1994. ftp://ftp.apnic.net/apnic/docs/historic/project-final-report.txt

^[45] An issue beyond the scope of this article.

^[46] “Asia Pacific Network Information Center Pilot Project Final Report (APNIC-009)”, 1 July 1994.

^[47] Minutes of CCIRN meeting, 21 June 1994, Amsterdam, The Netherlands. http://ccirn.net/general/minutes-Jun94.html

^[48] Conrad, D., “The Asia Pacific Network Information Center, Present and Future”, 1994, Connect.

^[49] ibid.

^[50] ibid.

^[51] “Asia Pacific Network Information Center Pilot Project Final Report (APNIC-009)”, 1 July 1994. ftp://ftp.apnic.net/apnic/docs/historic/project-final-report.txt

Note, this report also contains an appendix listing the membership of all the APNIC mailing lists.

^[52] Announcement of first APNIC meeting, late 1994. ftp://ftp.apnic.net/apnic/meetings/Jan95/announce

^[53] “IP Network Address Application Form Supporting Notes (APNIC-001)”, 16 February 1995. ftp://ftp.apnic.net/apnic/archive/apnic-001-v002.txt

^[54] Conrad, D., “APNIC FOO (Funding, Organization and Operation)” presentation, January 1995. ftp://ftp.apnic.net/apnic/meetings/Jan95/foom.ps

^[55] ibid.

^[56] “The APNIC Funding Plan and Current Status (APNIC-013)”, 1 September 1995. ftp://ftp.apnic.net/apnic/docs/historic/apnic-funding-status.txt

^[57] “Asia Pacific Network Information Center Organizational Details (APNIC-012)”

^[58] ibid.

^[59] Minutes of RIPE 20 meeting, January 1995, Amsterdam, The Netherlands. http://www.ripe.net/ripe/meetings/ripe-meetings/ripe-20

^[60] In 1994, APCCIRN changed its name to Asia Pacific Networking Group (APNG). https://sites.google.com/site/internethistoryasia/book2/section-1-2-apccirn-apng

^[61] “Asia Pacific Network Information Center Organizational Details (APNIC-012)”, 16 February 1995. ftp://ftp.apnic.net/apnic/docs/historic/apnic-org-details.txt

^[62] Unfortunately, most official documentation from APNIC 2 appears to be lost. The only surviving document in the APNIC archive is a list of attendees. ftp://ftp.apnic.net/apnic/meetings/Jul95/attendees

^[63] Private correspondence, 24 November 2012.

^[64] “The APNIC Funding Plan and Current Status (APNIC-013)”, 1 September 1995.

^[65] ibid.

^[66] “APNIC Status Report”, Third APNIC Meeting, Singapore, January 1996. ftp://ftp.apnic.net/apnic/meetings/Jan96/activities.ps

^[67] APRICOT website, February 2012, http://www.apricot.net/apricot96/

^[68] Unfortunately, it is beyond the scope of this article to document details of the many contributions to APNIC’s early history. In private correspondence – while admitting the list is far from comprehensive – David Conrad has called particular attention to the work of: Adam Peake, Andy Linton, Barry Greene, Che Hoo Cheng, Erin Chen (Yu Hsuan Chen), Geoff Huston, Gopi Garge, Haruhisa Ishida, Hiroyuki Fukase, Izumi Aizu, Jin Ho Hur, John Houlker, Jun Murai, Kanchanna Kanchannasut, Kilnam Chon, Koichi Suzuki, Kyoko Day, Laina Raveendran Greene, Masaki Hirabaru, Masaya Nakayama, Mathias Koerber, Osamu Nakamura, Patrick Kelly, Pindar Wong, Qian Hualin, Rahmat Samik-Ibrahim, Robin Erskine, Sanjaya, Shigeki Goto, Suguru Yamaguchi, Taeha Park, Tan Tin Wee, Tommy Chen, Toru Takahashi, Toshifumi Matsumoto, Toshiya Asaba, Vince Gebes, Xing Li, Yoshiko Chong.

^[69] Private correspondence, 12 November 2012.