Summary

  • The System Development Network (SDN) connected Seoul National University and a research institute in Gumi over a 1,200 bps leased line on 15 May 1982; the computers used TCP/IP.
  • That working domestic link did not make overseas exchange a continuous IP route. Early international mail and Usenet used UUCP and CSNET paths, and some Usenet material came on magnetic tape by post.
  • The direct KAIST–University of Hawaii line in 1990 and KT’s commercial KORNET service in 1994 were later, different milestones.

A line connected two research computers

The date most often repeated for Korea’s Internet beginning is unusually specific: 15 May 1982. In Kilnam Chon’s retrospective, a PDP-11 at Seoul National University was linked to a VAX-11/780 at a research institute in Gumi over a dedicated 1,200 bits-per-second line. The institute is described by its later name, the Electronics and Telecommunications Research Institute (ETRI). ETRI’s own anniversary account gives the same date, machines and line speed.

The project was called the System Development Network, or SDN. Its purpose was to test computer-networking techniques and improve technical information exchange. The computers spoke TCP/IP. In January 1983, a VAX at KAIST joined the network, adding a third host. These details describe a real domestic research network, but they do not measure how many people could use it, how much traffic it carried, or whether every institution in Korea could reach it.

Later national accounts sometimes call the 1982 link the country’s first Internet connection and even place Korea second in the world. The evidence here supports a narrower statement: it was an early, working TCP/IP connection inside Korea. It does not supply a common definition or a comparative record for the international “second country” claim.

Overseas exchange followed other paths

The cross-border story did not simply extend the leased line. Chon’s account says SDN used UUCP in August 1983 to reach EUnet through MCVAX in the Netherlands. That October, a connection through HPLABS brought SDN into UUCPNet and enabled email exchange. These were store-and-forward paths: a domestic TCP/IP link and an international exchange mechanism could coexist without being one continuous IP service.

In December 1984, SDN connected to CSNET through a service provided by the U.S. Geological Survey. Chon says the link became an important channel for technical exchange before Korea connected directly to the Internet. U.S. access restrictions limited what could be used: email and Usenet were available, while FTP was not. High international telephone costs also shaped the service. Usenet material could be copied to magnetic tape and sent by post instead of being fetched over an expensive line.

That tape is not a quaint footnote to a fully connected network. It marks the difference between information that crossed a border and a route that an operator could use interactively. A message, a news spool, and a bidirectional IP path were different services with different dependencies.

Direct IP and public service came later

The National Information Society Agency’s history dates the formation of SDN/HANA to 1989 and Korea’s direct Internet connection to 1990, when KAIST and the University of Hawaii received a dedicated line. ETRI and Korean government retrospectives place commercial Internet service later: Korea Telecom’s KORNET began in 1994, alongside services from other providers.

The chronology therefore has several valid beginnings. A domestic TCP/IP network operated in 1982. UUCP and CSNET carried selected international services in the following years. A direct international IP line came in 1990. Commercial Internet service followed in 1994. Calling all four “the first connection” would erase who could use the system, what it carried, and which route actually existed.

The detailed route and tape account comes mainly from Chon’s 2011 retrospective; the cited histories are not packet traces or line contracts. They do not quantify traffic, users, costs or reliability. New operating records could sharpen the dates and service boundaries. Until then, the safest history is the one that keeps domestic TCP/IP, international message exchange, direct IP and public service separate.

Sources