Summary

  • Kilnam Chon’s SDN began with two computers, a 1,200 bps line between Seoul National University and KIET in Gumi, and a software implementation of TCP/IP routing made from standards documents.
  • The network expanded by combining unlike mechanisms: domestic TCP/IP, international UUCP, Korean-language mail, shared financing, address and domain administration, and eventually a transfer of operations beyond the university laboratory.
  • Open specifications reduced dependence on unavailable equipment, but the record credits a wide operating community and does not show that one router or one person caused South Korea’s later broadband success.

The appliance that was not there

The first SDN topology can be described without grandeur. A computer in Seoul National University’s Department of Computer Science was connected to one at the Korea Institute of Electronics Technology in Gumi. The line ran at 1,200 bits per second. Kilnam Chon and his co-authors date the start of operations to 15 May 1982; a KAIST computer became the third node in January 1983.

The interesting constraint sat between the endpoints. APNIC’s later oral history says Chon and his students could not obtain a router from ARPANET and that commercial IP routers were still about five years away. They developed routing in software by working from standards documents. That account should not be mistaken for a surviving procurement file or a complete catalogue of the early router market. It is retrospective testimony. Yet it identifies an important difference between an open protocol and a finished product: the protocol can be implemented where the product cannot be acquired.

That choice did not eliminate cost or risk. It relocated them. The team had to interpret specifications, make unlike computers interoperate and diagnose faults without a vendor’s appliance or support contract. A standard was an option to act, not a warranty that the action would work. SDN became meaningful because people accepted that implementation burden.

One network, several kinds of connection

Calling SDN an Internet can blur how uneven its edges were. TCP/IP operated among its computers, but its first international reach used other tools. The co-authored history records a UUCP connection to mcvax in the Netherlands in August 1983 and another to hplabs in the United States that October. In December 1984, SDN connected to CSNET.

Those links did not instantly offer the full range of Internet services. US restrictions on access to ARPANET meant that email and USENET news were available with the United States while services such as FTP were not. International telephone charges made the data path stranger still: a large portion of USENET material arrived on magnetic tape by post. A network could therefore be live, international and useful while combining packet links, scheduled transfers and a physical mailbag.

AsiaNet extended this pragmatic model among sites in Australia, Indonesia, Japan, Korea and Singapore, with additional links to Europe and North America. It would be inaccurate to recast all of those paths as end-to-end TCP/IP. The formal 56 Kbps link from KAIST to the University of Hawaii came only in March 1990. Distinguishing the layers matters because “connected” described different service boundaries at different dates.

A Korean message was an infrastructure test

Reach was not the same as usability. A network whose mail path damaged Korean characters could connect machines while excluding ordinary Korean expression. The technical history records a KAIST master’s thesis on a Korean-character mail system in 1983, experiments that followed, and a Korean mail program plus the hvi editor in 1985.

The work later became part of a public technical record. RFC 1557, issued in December 1993 by U. Choi, K. Chon and H. Park, describes ISO-2022-KR for Korean characters in Internet messages. It says the method had been specified in 1991 and was widely used in Korean IP networks. The RFC is Informational, not standards-track, and Chon was one of three authors. Its significance here is narrower and more practical: local use generated a format that other implementers could inspect.

The same sequence complicates a hardware-centred origin story. The early router allowed packets to move. Korean-language tools helped make the resulting service worth using. One solved an unavailable appliance; the other solved an unavailable representation. Both required local engineering beyond attaching a line.

Naming, payment and handoff

By 1985, APNIC’s account says SDN linked around twenty universities, national research laboratories and corporate laboratories, alongside overseas sites. Scale created administrative work that could not remain an informal property of the first nodes. Korea received the public IPv4 block 128.134.0.0 in July 1986. Rules beneath .kr were established, and the country-code domain was operating that year.

The Academic Network Committee formed in 1988 to coordinate domestic address and domain assignments, overseas connections and international representation. This was not merely bureaucracy added after the technical work. Names and numbers decide who can be reached and recognised. Moving those decisions into a defined committee reduced dependence on the original laboratory while creating a new surface whose membership and accountability mattered.

International capacity exposed another boundary. Many SDN institutions jointly financed a 56 Kbps leased line to Hawaii through HANA. When the satellite link joined KAIST to the University of Hawaii in March 1990, HANAnet offered far less constrained global use than the metered UUCP and CSNET arrangements. Shared financing crossed a threshold no single laboratory was assumed to carry alone.

Operations then moved. In August 1992, the principal gateway equipment and operation of HANAnet and SDN transferred from KAIST to Korea Telecom. The co-authored history connects HANAnet to the later emergence of KORNET, KT’s commercial Internet service. That is evidence of a handoff, not evidence that the research network single-handedly produced a commercial market. The distinction keeps credit with the institutions that funded, operated and expanded each stage.

The human network outgrew the first link

Chon’s role continued through convening. The 1985 Pacific Computer Communications Symposium in Seoul brought together approximately 300 participants from Asia, Europe and North America, according to the technical history. Later recurring workshops expanded the exchange of operational knowledge. APNIC’s institutional biography also associates Chon with the formation of APNG, APAN and APTLD.

At the first formal APNG meeting in 1991, Chon later recalled, participants approved the need for a regional Internet registry. The path to APNIC remained collective. APNIC’s history project describes a 1992 CCIRN meeting in Tokyo and a September 1993 experiment distributed across organisations: JPNIC supplied budget, space and hardware; KRNIC supplied information services; AUNIC supplied DNS-related services; a mailing list of twenty-five people across several economies served as staff.

This later arrangement echoes the first software router without simply repeating it. Both divided a missing capability into work that available participants could perform. But an institution is not software. Its mandate, funding and accountability cannot be compiled from a protocol document. Chon’s contribution is best understood at that boundary: he helped demonstrate technical possibility, trained people to carry it, and joined organisations that made regional coordination less dependent on any single node.

The Internet Society gave Chon the Jonathan B. Postel Service Award in 2011 and inducted him into the Internet Hall of Fame’s inaugural class in 2012. Those honours support the record of service. They do not turn a network of students, laboratories, operators, funders and regional peers into one man’s invention.

Sources