Fujisawa

The next morning I was up early, having heard horror stories of how hard it would be to navigate the succession of subways and trains in the middle of rush hour.

Yes, there were packers to get more people into the subways. Since the density was only about equivalent to a crowd leaving a football game, however, the packers weren’t yet needed. They sat watchfully by, wearing white gloves and ready to jump into action when the time came.

The density of the crowds was amazing. Passing the Tokyo suburbs, I expected the crowds to disperse, but the Yokohama commute brought even more people onto the trains. Even outside of Yokohama in Fujisawa, the density remained constant.

Arriving at Tsujido station, I waded through huge crowds until I got outside the station. There, I started waving Jun Murai’s card around and mispronouncing “Keio,” my destination, until a kindly old gentleman in an indeterminate uniform pointed me to the stop for bus number zero.

I queued up and boarded. We passed through Fujisawa and started to head into the countryside. Throughout the ride, every time we passed a cluster of university-like buildings, I would accost some hapless passenger, mumble “Keio,” and wave the business card around. One by one, the passengers waved for me to sit down.

Finally, we reached the end of the line. Rising out of the countryside, sitting on a hill, was a futuristic clump of buildings, surrounded by empty fields. I made my way up a long set of stairs into the central courtyard. In marked contrast to the subways, trains, and streets, constantly teeming with people, the Fujisawa campus of Keio University, Japan’s best private university, was absolutely deserted.

Feeling like I was in some futuristic ghost town, I wandered around random buildings until I chanced upon a group of administrators hidden away in a back office. Waving Jun’s business card produced a flurry of whispered consultations until a map of the campus suddenly appeared and I was pointed towards building number zero.

Going to the third floor, I knew I was in the right place when I saw an office with a pile of empty computer boxes piled outside. I parked myself on the boxes and waited for Jun. It was an eerie feeling, sitting there in this deserted building on this deserted campus listening to the building creak and groan. Occasionally, a door would open and a person would scoot out of a room and scamper away, leaving only the echoes of their footsteps.

I was to find later that the deserted campus was merely an illusion. Being a self-employed idler, I had arrived well past 9, by which time everybody was already hard at work.

Jun Murai’s secretary soon arrived. Known worldwide as “Jun sec,” her e-mail address appears on all WIDE literature, posters, stickers, and the like. Junsec let me onto an X terminal to read the several hundred mail messages that had accumulated since Honolulu. Most of the mail was about Bruno, the standards server.

Evidently, word had gotten out. Bruno was running 24 hours a day, with load averages of 35 packets per second not uncommon. So many mail requests had come into the infoserver software that the batch queue had over 150 unfulfilled jobs. The amount of FIP traffic was so heavy that we had become a serious user of bandwidth on the transatlantic links.

This posed a delicate problem. On the one hand, we wanted lots of people to use the server in Colorado so that we could prove to the ITU that the service was needed. On the other hand, being good network citizens meant setting up mirrored servers, putting replicas of the data on other machines to avoid having too much redundant traffic cross the backbone.

Offers to provide mirrored servers were pouring in from all over the world. Comparing notes with Tony Rutkowski, it appeared that every country in Europe had volunteered to maintain a set of the standards. Several hundred hosts had already received files from Bruno and it looked like we were getting FIP traffic from over 25 countries.

When I was almost done plowing through my mail, Jun Murai came bursting in. Disheveled, animated, and wearing his trademark blue jeans, he greeted me enthusiastically. Taking me into the obligatory conference room, we began talking while Junsec brought in cups of tea (she had already brought me two cups of coffee).

I presented Jun with a copy of Stacks for his boss, Dean Aiso. He presented me with a handsomely bound copy of the WIDE annual report, several hundred pages of kanji with gold-embossed lettering on the cover. My manners being better than my Japanese, I thanked him while mentally trying to figure out how to get this heavy and, to me at least, cryptic tome back to the States.

Formalities duly dispensed with, we started talking about WIDE. Jun is only 36 years old and I was curious how, in a stratified society like Japan, somebody like him had ended up running the Japanese Internet.

In 1984, NTT still had a throttle on all telecommunications. Putting alien devices (e.g., modems) on the telephone lines would be considered about as proper as greeting the Emperor by slapping him on the back. However, it was widely known that in April, 1985, NTT would deregulate.

All the senior researchers had been debating how to take advantage of deregulation to put in networks. Meetings were held to debate the subtleties of various OSI architectures. To Jun, this was a waste of time. As he puts it, “I was young and that was boring.”

He took two modems, scammed a phone line from university administrators (no easy feat), and started running UUCP transfers. That was the start of JUNET. While the establishment continued to attend OSI meetings, JUNET continued to grow. Links were set up to a machine named mcvax in Amsterdam (the precursor to EUnet) and to seismo in Washington, D.C. (the precursor to UUnet). By 1986, a domestic IP network had started and by 1989, Jun Murai and Torben Nielsen had established a link to Hawaii. Larry Landweber helped to hook up Japan to CSNET.

Meanwhile, the powers that be continued to debate OSI. When they finally looked up from their deliberations, Jun already had several hundred nodes on his network. As in much of the world, while committees waited for OSI, a few people turned TCP/IP networks into a reality. Japan, like other countries, had many people trying to legislate networks into existence while a few people rolled up their sleeves and installed cable.

By the 1991 school year, WIDE had grown to the point where it had an annual budget of U.S. $1 million raised totally by donations. WIDE was officially a research project, involving 57 researchers, roughly half from the university community. The operational requirements of WIDE became so demanding that Jun was actually turning away requests to be connected until a suitable infrastructure could be set up to run the net.

During our conversation, Jun repeatedly referred to a desire to stop running networks and get back to his real research. Even with the demands of running WIDE, he has compiled an impressive record.

When JUNET was first being put into place, Jun noticed “we had a network but nobody was using it.” Networks are like a fine dinner: the whole effort is wasted if nobody comes. The reason in the case of JUNET was quite simply that e-mail and USENET used the Roman alphabet and Japan uses kanji characters. Jun changed that. He and others added kanji support to the X Window System, kanji character handling for RFC 822 mail, and multibyte character handling for the C programming language.

He also helped design a font server. This software lets users start spelling out characters in romanized Japanese. As the words begin to be formed, Kanji characters start appearing on the screen. When the right character shows up, the user points to it and proceeds to the next word.

When I visited Fujisawa, Jun was involved in fascinating projects to support mobile hosts and mobile people. One of the more interesting projects was the Phone Shell. The Phone Shell is a fully functional UNIX shell which takes input from a 12-key telephone pad instead of a keyboard or a mouse. Typically, users execute scripts, such as having mail messages redirected into a voice synthesizer.

As Jun explains it, “I can go to the bar and drink beer. I go to a phone and ping my routers, and if they are still working, I go back and drink more beer.”

Another project brought ISDN into the TCP/IP protocol suite. Unlike other countries, which are still demonstrating the viability of demos, Japan has deployed ISDN as an operational service. In fact, WIDE uses ISDN lines to supplement leased lines in case of congestion or failure.

The WIDE ISDN module is fully integrated into TCP/IP. When a datagram for an ISDN-reachable source is encountered, a call is placed. The delay to set up the circuit ranges from 800 ms to 3 seconds. Once established, the line stays up until it has been idle for a user-configurable period, at which point it is taken down.

ISDN can be used for more than just routers, however. Jun has TCP/IP on his laptop. Tokyo has ISDN pay phones. He can bring a laptop into a phone booth and be a fully functioning member of the Internet. Even Superman would be jealous.

Another project close to operation when I visited was the use of satellite circuits for home PCs. Japan has deployed satellites as an alternative to cable TV. Dishes cost as little as U.S. $100 each. Satellites for TV signals are a one-way data channel, but they operate at high bandwidth. A PC with an ISDN card can use 64-kbps “B” channels to send commands. Data coming back to the PC can use up to the 8 Mbps wide-area bandwidth of the satellite dish.

Ethernet cards are used to give the PC a 10 Mbps interface in a LAN environment. Jun and his staff had linked the Ethernet card to the satellite receiver to give the home user the ability to do WAN based multimedia, large file transfers, and other operations requiring large amounts of bandwidth.

We went from Jun’s laboratories to tour the campus. Fujisawa was a brand new campus of Keio University. In sharp contrast to other Japanese (and most U.S.) universities, senior researchers at Keio teach freshmen and sophomores. Jun delighted in telling how he would instruct a class of freshmen, “now we will ping Switzerland.”

At the Fujisawa campus, all entering freshmen are required to learn UNIX, to operate workstations, and to acquire other basic skills. Although laptops are not required, they are very strongly encouraged.

We walked into the library where students run an ID card through a reader to enter. Of course, Jun forgot his card, so we crawled over the barriers. We walked into a lounge area where groups of students sat clustered around HDTV sets watching assignments.

Around them, students were at workstations, each equipped with a cassette tape drive and a headphone, doing language exercises. Other students were debugging programs.

Finally, it dawned on me.

“Jun, there are no books in here.”

“You still have books?” he asked with a smile. The books were up one floor, but it still gave the library a strange feeling.

Walking outside, we spotted Junsec patiently waiting for us in a car. We all went into Fujisawa where we were ushered into a private tatami room for an exquisite lunch of many courses, each beautifully arranged and delightfully prepared.

As I was gulping down the tenth course, a delicate broth with a single tiny mushroom and a little piece of fish, Jun started explaining to me how the spindly little mushroom I had just inhaled was handpicked in the forests and only available for a few weeks each year. Each mushroom cost as much as 5,000 yen (U.S. $35). I puffed out my cheeks as if I were still rolling the mushroom in my mouth to savor the delicate flavor, while Jun went on to explain how rare the fish was.