Carl Malamud: Internet Talk Radio, flame of the Internet.

Malamud: This is Geek or the Week, and we’re proud to have the ulti­mate geek here today. This is Jun Murai from Keio University. Jun, wel­come to Geek or the Week. I’d like to ask you about some of the work you’ve been doing in Japan. Maybe you can start by telling us about the work you do at Keio uni­ver­si­ty. You have stu­dents there that need to com­pute, right?

Murai: Right. Okay. Our uni­ver­si­ty, Keio University has eight cam­pus­es. And this is a pret­ty new cam­pus, just four years old. So we’re hav­ing the first grad­u­a­tion in next March. So each of the grades has about 1,000 stu­dents. So we have 4,000 stu­dents now. And we’re going to have a grad­u­ate school. But this uni­ver­si­ty is very spe­cial for the Japanese stu­dents and Japanese uni­ver­si­ties, and maybe for uni­ver­si­ties world­wide. When they get into the uni­ver­si­ty, they have to learn how to use the Internet. Every sin­gle stu­dent, right. And then they have to use email and oth­er net­work­ing func­tions on Unix work­sta­tions for their entire cam­pus life.

Malamud: So these are fresh­men, they come in, they need to learn Unix, they need to learn Internet skills, that’s a part of their education.

Murai: Right. That’s the first thing they have to do.

Malamud: And how do they do that? Do you have work­sta­tions all over cam­pus that they sit down and use?

Murai: Yeah, that’s the point. Yeah, we want­ed a lot of work­sta­tion for them, but you know, it was impos­si­ble because of the finan­cial prob­lem we had, and also we did­n’t have the space. So then we were ask­ing all the Japanese com­put­er mak­ers to pro­vide us with a very light Unix lap­top com­put­er. And they worked very hard but they could­n’t make it at that time, four years ago. And then Sony pro­duced the Soney NEWS work­sta­tion, which is a Unix work­sta­tion, and also there was a SpArc lap­top, the first one, from Toshiba. But those are very expen­sive and very heavy, and it was impos­si­ble for students.

So, what we thought was to pro­vide very light lap­tops to the stu­dents, so that they can use their work­sta­tion but if the num­ber of work­sta­tions is not enough then they can use the lap­top in exact­ly the same man­ner. That’s was what we want­ed. So, what did gave up on the Unix work­sta­tion lap­top solu­tion. So instead we chose a DOS lap­top up Toshiba for the first year, and then we start­ed all the stu­dents hav­ing a very sim­i­lar envi­ron­ment on a DOS lap­top. And then in the uni­ver­si­ties, prob­a­bly one third of the stu­dents could use Unix work­sta­tions. So that was how we started.

Malamud: And are they still run­ning DOS lap­tops and Unix work­sta­tions now?

Murai: Yes. That’s a very good ques­tion, actu­al­ly. Okay. So now we’ve been work­ing on real­iz­ing our dream. So we have been work­ing on mak­ing the lap­tops run Unix. And then final­ly we got that kind of envi­ron­ment. So we actually…probably last year, we got a copy of BSD run­ning on the 386 archi­tec­ture. And then we put the all the envi­ron­ments we’re using at the uni­ver­si­ties, which is most­ly hand-made and pub­lic domain Japanese character-handling envi­ron­ment… So we put all those envi­ron­ments on that, so we are now run­ning machines.

Malamud: Now how do these things get into the net­work then? I can under­stand with the work­sta­tion, but you just have a bunch of dock­ing ports that peo­ple get into? Do you have wires all over campus?

Murai: Yeah, that’s on the prob­lems, too. So of course we are work­ing with an Ethernet solu­tion so that lap­tops can con­nect to the uni­ver­si­ty cam­pus net­work via nor­mal Ethernet like 10BASE‑T and 10BASE2, because our cam­pus was designed that way from the first point. So every sin­gle room is equipped with Ethernet anyway. 

But then we hit the prob­lem that know, because all 4,000 stu­dents are car­ry­ing their lap­top and try­ing to con­nect to the cam­pus net­work some­where in some room or…then we need the wires and the cables every­where, right. We can’t do that, so we are now work­ing on a wire solu­tion as well. But now that we have Unix run­ning on it, that’s a nor­mal Unix, so we can devel­op any kind things over it. 

And also we are now try­ing to put— Okay. Since we have build­ings, right, and each of the build­ings are on a sep­a­rate sub­net. So if a stu­dent is car­ry­ing a lap­top from one build­ing to the next, then the sub­net address would change. So the WIDE Project, our group, is work­ing on a mobile IP pro­to­col. So we are now up putting that pro­to­col onto the lap­top Unix so that they can move around the build­ings, and hope­ful­ly the wire­less solu­tion will be com­plet­ed by March.

Malamud: So there’s two aspects. One is the wire­less com­mu­ni­ca­tions, the oth­er is the mobile IP.

Murai: Right. Yeah. [crosstalk] Mobile—

Malamud: How do you do wireless?

Murai: Mobile came first, and we’re going to work on the wireless.

Malamud: Okay, wire­less is just going to be some infrared or microwave, or—

Murai: Yeah— Well… Yeah, cur­rent­ly we are think­ing about using the tele­com wireless.

Malamud: Now what about the mobile IP prob­lem? I know a lot of peo­ple have been work­ing on it. It sounds like you have solu­tion for mobile like IP that’s in the works.

Murai: Well actu­al­ly the one solu­tion pro­posed for IETF is from a Sony [indis­tinct phrase] work­ing on that. He’s very active­ly work­ing with­in the WIDE Project.

Malamud: Now how does that work? What hap­pens when I move from— I’m in one sub­net and I’m com­put­ing, and I pick up my lap­top and I walk into anoth­er build­ing. What hap­pens? Is there some dynam­ic read­dress­ing going on? Uh…

Murai: Uh…yeah. Actually the… Yeah, there are solu­tions for the var­i­ous cas­es, but the solu­tion we are employ­ing is called VIP—Virtual IP. So basi­cal­ly one machine’s got two types of IP address­es. And then an inter­me­di­ate node, either…kind of switch­ing the oth­er for the, you know, to assure the…you know.

Malamud: So it’s a lev­el of indi­rec­tion, is that what we’re doing here?

Murai: Right.

Malamud: Okay. And that means that as I walk from build­ing to build­ing, things like my TCP con­nec­tions are not bro­ken, right?

Murai: That’s true. 

Malamud: I just keep on going.

Murai: Yeah, that’s true. And even the— You know, we have a lot of exper­i­ments of the vari­a­tions on the imple­men­ta­tion of the pro­to­col. So even the TCP connection…while using the TCP con­nec­tion and mov­ing, or cut­ting, and you know, switch­ing from Ethernet to a SLIP con­nec­tion, you still keep the TCP con­nec­tion, so…

Malamud: That’s amaz­ing. So that’s… Walking is fair­ly slow. Now, is this gonna work if I’m dri­ving in my car? Is this robust enough to han­dle a car situation?

Murai: Uh…well, yeah. Yeah, I’m sure it… Okay the VIP solu­tion is inde­pen­dent from the lower-level solu­tion like wire­less things. So that’s an inter­est­ing point we are doing in our cam­pus because we’re going to use the wire­less thing andthe mobile VIP solu­tion at the same time, right. So you know, car­ry­ing the lap­top from class­room to class­room we’re using the lower-level solu­tion of the wire­less net­work­ing tech­nol­o­gy. And then if you jump from build­ing to anoth­er, then the mobile IP solu­tion would help you. 

Malamud: Interesting. Now, you run…in addi­tion to doing work at Keio, you’re the founder of WIDE, which is the Japanese Internet. Give us some idea of how big this net­work is.

Murai: Okay.

Malamud: Are talkin’ a dozen com­put­ers? A cou­ple dozen?

Murai: Okay. We have seven…a lit­tle bit more than a dozen, net­work oper­a­tion cen­ters all over Japan, from the north­ern island Hokkaido to the south­ern islands. So sev­en loca­tions, and we have a so-called back­bone. It’s not very big band­width, though. So that entire net­work is used for the exper­i­men­tal work of the devel­op­ment of the pro­to­cols and you know, that kind of thing. But at the same time we are pro­vid­ing the Internet ser­vices to each of the mem­bers of the WIDE Project. The Wide Project is to real­ize the widely-integrated dis­trib­uted envi­ron­ment, that is what ​“WIDE” stands for. To real­ize a glob­al [indis­tinct] envi­ron­ment which is very sim­i­lar to—

Malamud: How many peo­ple are part of the WIDE Project?

Murai: Okay, the reg­is­tered researchers work­ing for the WIDE Project is 130. No, 136 mem… [crosstalk] researchers and students.

Malamud: So 136 researchers. I mean we’re talkin…—

Murai: Yeah, researchers and stu­dents. But active mem­bers are about eighty-six or…yeah.

Malamud: And how many com­put­ers do you esti­mate are in area served by WIDE? [crosstalk] How big is the Japanese Internet?

Murai: Well… Oh how big is the Japanese Internet. Okay. About… Let’s see, about 500 orga­ni­za­tions con­nect­ed? Network num­bers con­nect­ed, sorry.

Malamud: Mm hm.

Murai: Yeah. So—

Malamud: So that’s 500 net­works, and pre­sum­ably many thou­sands of com­put­ers, or tens of thou­sands of computers.

Murai: Well…yeah, not. Yeah, there are a lot of a small enti­ties. But yeah, there are very big enti­ties as well.

Malamud: Now, WIDE orig­i­nal­ly start­ed out as a research net­work, as a research exper­i­ment in fact.

Murai: Exactly.

Malamud: And now it looks like the Internet in Japan is start­ing to enter a new phase. You’ve been instru­men­tal in help­ing to give birth to IIJ, which is a…service provider?

Murai: Yeah. Okay. So the prob­lem in Japan was… We’ve been work­ing on pro­mot­ing this kind of envi­ron­ment like Internet in Japan. And that has been very suc­cess­ful and every­body under­stands that okay, the Internet is a good thing and that this kind of com­put­er net­work is very use­ful. And now every­body wants to get on the Internet. That was what I want­ed.

Malamud: Jun Murai, let’s talk a bit about your research. You actu­al­ly are a researcher in addi­tion to run­ning a large net­work. There’s a cou­ple projects you’ve been involved in that I find I absolute­ly fas­ci­nat­ing. Maybe we could start by talk­ing about satellite-based mul­ti­cas­t­ing. Could you tell us what that project is?

Murai: Ah yes. Okay. When we are run­ning WIDE Internet, we found that there are a lot of appli­ca­tions over the WIDE Internet were basi­cal­ly the mul­ti­cast com­mu­ni­ca­tion. So we were inter­est­ed in mul­ti­cast com­mu­ni­ca­tion tech­nol­o­gy. And so we worked on a Class D IP trans­mis­sion thing. 

Now, at the same time we are are a research net­work and so we are fund­ed by var­i­ous com­pa­nies. But for the research, we are not rich at all. So we are always look­ing for more band­width. And then one file was shared by every­body, and the net­work some­times con­gest­ed. So then when you look up the [indis­tinct] there is a satel­lite broad­cast­ing start­ed. And I talked with them and there was cer­tain band­width that they weren’t using, and they want­ed that bound­aries for data com­mu­ni­ca­tion but they did­n’t know how. So we’ve been work­ing with them.

Malamud: Now, we’re talk­ing about the K band satel­lite sys­tem, right, in Japan?

Murai: Right.

Malamud: Now, K band for lis­ten­ers who don’t know, it con­sists of if you walk around Tokyo you will see lit­tle satel­lite dish­es in almost every home. And these dish­es cost a hun­dred bucks, right?

Murai: Right. Yeah. That’s very cheap. And that is a point we’ve been work­ing for. And then okay, satel­lites. We’re gonna have band­width. And we can design the mul­ti­cast trans­mis­sion. But if there’s just one too many just using the satel­lite, that is not help­ing the Internet world at all. So what I was work­ing on was inte­grat­ing the satel­lite com­mu­ni­ca­tion paths togeth­er with the exist­ing ter­res­tri­al com­mu­ni­ca­tion path, which is to use the satel­lite com­mu­ni­ca­tion path as part of the nor­mal Internet com­mu­ni­ca­tion path, right. 

So then we designed a spe­cial kind of rout­ing mech­a­nism so that if you issue a mul­ti­cast data­gram, then it would be auto­mat­i­cal­ly rout­ed to use a satel­lite. The mech­a­nism is basi­cal­ly very sim­i­lar to the [force?] rout­ing type of thing, but it’s a mix­ture of the exist­ing tech­nol­o­gy plus some new ideas. 

But any­way. So then as you men­tioned, there is a satel­lite mar­ket around in Japan now. And so you can get a receiv­er in a very… Yeah, it’s kind of very cheap, right. 

Malamud: But it con­nects to a TV nor­mal­ly, right? [chuck­les]

Murai: Right. Yeah. So what we did was we designed a small hard­ware adapter which can retrieve infor­ma­tion from the nor­mal receiv­er, and then pro­vide that data to the com­put­er, right. It’s an adapter; we call it a satel­lite adapter. We designed the hard­ware. And the inter­est­ing por­tion of the hard­ware is then— Okay. Actually, poten­tial­ly the satel­lite could car­ry 8 megabits per sec­ond, right. Then if you think about the inter­face board, right, to receive 8 megabits of ser­i­al line com­mu­ni­ca­tion, then usu­al­ly you have to buy a spe­cial kind of ser­i­al line board, right. Which is very expen­sive usu­al­ly. So if we do start this tech­nol­o­gy for the actu­al com­mu­ni­ca­tion, prob­a­bly the uni­ver­si­ties and peo­ple who can buy that board are you know, prob­a­bly lim­it­ed. And also the prod­ucts are lim­it­ed. So for your com­put­er receive the satel­lite com­mu­ni­ca­tion would be lim­it­ed. And that is not inter­est­ing at all. So that is not very good.

So what we designed was an 8 megabit [indis­tinct], and then a cheap inter­face. Then the only down­load com­mu­ni­ca­tion would be used for that path. Okay. Then why can’t we use Ethernet? Then if that is emu­lat­ing the Ethernet inter­face, then the inter­face board is very cheap and we can get— Okay, that Ethernet inter­face just pro­vid­ing the satel­lite datas­tream, so noth­ing else. So in that case we can get prob­a­bly very good per­for­mance from that board, that adapter. So we decid­ed to emu­late the Ethernet inter­face from the satel­lite adapter. And we about thir­ty pilot hard­wares now, and we are run­ning, and we are test­ing, and it’s run­ning pret­ty well.

Malamud: So for a few hun­dred dol­lars for a K band satel­lite dish, which is cheap, an Ethernet card, which is cheap, and this satel­lite adapter that sits in the mid­dle, you give your users 8 megabits of wide-area communications.

Murai: That’s true, yeah.

Murai: Class D. Currently Class D is rout­ed to the satel­lite com­mu­ni­ca­tions. And then on the way back, like an ACK type of com­mu­ni­ca­tion, is now using ter­res­tri­al. So it’s a com­bi­na­tion of satel­lite one way, and then ter­res­tri­al on the oth­er way. So the oth­er way it’s nor­mal IP rout­ing. It’s some­thing with an assump­tion that your machine has a receiv­er, then you get the ben­e­fit from the satel­lite com­mu­ni­ca­tion which is you know, a broad­cast and mul­ti­cast type of thing. And at the same time you have to a ter­res­tri­al IP link the WIDE Internet right now.

Malamud: And is that ISDN? Is that a [crosstalk] fast modem? 

Murai: Any. Any kind of— We have devel­oped IP over ISDN thing for the first stages of our project. So and any kind of—

Malamud: Now, you make exten­sive use of ISDN in WIDE, don’t you?

Murai: Right. 

Malamud: That’s an impor­tant tech­nol­o­gy for you.

Murai: Yeah, right. In our coun­try. And you know, the Japanese gov­ern­ment is turn­ing to start up the BISDN, FTTH in a pret­ty— Probably by the end of the century.

Malamud: But reg­u­lar ISDN is out there, it’s just avail­able. You go any place in Japan and you [crosstalk] can buy ISDN lines?

Murai: Yeah, any place, any place. And you know—

Malamud: Pay phones, and…

Murai: Yeah. Pay phones, yea. That’s an inter­est­ing por­tion. So if you vis­it Japan— Actually you did. But there you’ll see phones with an RJ-45 jack. So if you have a lap­top equipped with an ISDN port, you can get prob­a­bly 128 Kbps on a street cor­ner. So you can [indis­tinct] from a street corner. 

Malamud: Which you do on a reg­u­lar basis.

Murai: Right. 

Malamud: Well that’s fas­ci­nat­ing. One of the oth­er projects you’ve been work­ing on is some­thing that you call XITV. What’s XITV?

Murai: Yeah that’s a new one, and actu­al­ly we are prepar­ing for that, and then we have some demon­stra­tions. It’s TV com­mu­ni­ca­tions, or video com­mu­ni­ca­tion, over the Internet [fame?]. But it’s cur­rent­ly a very sim­ple thing. The only pro­to­col the soft­ware is using is FTP. And—

Malamud: Now wait a minute. Let me just make sure I under­stand this right. Your trans­port lay­er for your television-over-the-Internet is…FTP.

Murai: FTP. Because we want­ed to use anony­mous FTP. Because— Okay. That is imple­ment­ed and cod­ed by Professor Yamamoto of Hokkaido University. So the basic idea of the soft­ware is very easy imple­men­ta­tion and han­dling on the client side. So basi­cal­ly if you have X Window run­ning, then you need noth­ing else, right. So what the soft­ware is doing is show­ing the video images to the X Window and pass­ing the voice data to the normal—.au file, actu­al­ly, but you know, if you have a Sun work­sta­tion then you can hear the voice con­tin­u­ous­ly. And then of course if it’s over the Internet, and espe­cial­ly the Japanese Internet is not that high-bandwidth, then the soft­ware is decreas­ing the num­ber of frames per second.

Malamud: But we’re talkin’ full-color, full-motion, if the band­width is there.

Murai: Well, full-color, full-motion, but the num­ber of frames is small. It depends on the con­di­tion of end-to-end link. And also there are cer­tain servers around, and they’re cap­tur­ing live video images and stor­ing that to disk, and then using the FTP pro­to­col retriev­ing both voice data and TV images. 

So that’s one. And then some­times the [indis­tinct] con­di­tions the link, but some­times only one frame per five sec­onds or that kind of con­di­tion would hap­pen. But then found out that that is good. I mean, nor­mal­ly good for the nor­mal broad­cast­ing of TV, like if you are watch­ing the TV news, and if you can hear the voice and if you can get one frame per three sec­onds, then you can prob­a­bly get most of the infor­ma­tion out of the TV broadcast. 

Malamud: Well that’s fas­ci­nat­ing. What oth­er things are you work­ing on now? 

Murai: A very impor­tant is han­dling glob­al mul­ti­lin­gual mes­sages on email mes­sages or Internet mes­sages in gen­er­al. We have a long his­to­ry from the start of the net­work of han­dling of local char­ac­ters, like know, kan­ji char­ac­ters, Japanese char­ac­ters. And that was a very impor­tant thing. And so we’ve just start­ed the trans­la­tion of the RFC doc­u­men­ta­tion into Japanese. And that will be very impor­tant for pro­mot­ing the tech­nol­o­gy of the Internet in our coun­try, right. So when we start­ed the email cul­ture in our coun­try, nobody was using it until we had devel­oped han­dling of local char­ac­ters on email and the relat­ed environments—operating sys­tems and the email soft­ware, and everything. 

Malamud: This is Internet Talk Radio, flame of the Internet. You’ve been lis­ten­ing to Geek of the Week. You may copy this pro­gram to any medi­um, and change the encod­ing, but may not alter the data or sell the con­tents. To pur­chase an audio cas­sette of this pro­gram, send mail to radio@​ora.​com.

Support for Geek of the Week comes from Sun Microsystems. Sun, The Network is the Computer. Support for Geek of the Week also comes from O’Reilly & Associates, pub­lish­ers of the Global Network Navigator, your online hyper­text mag­a­zine. For more infor­ma­tion, send mail to info@​gnn.​com. Network con­nec­tiv­i­ty for the Internet Multicasting Service is pro­vid­ed by MFS DataNet and by UUNET Technologies.

Executive Producer for Geek of the Week is Martin Lucas. Production Manager is James Roland. Rick Dunbar and Curtis Generous are the sysad­mins. This is Carl Malamud for the Internet Multicasting Service, town crier to the glob­al village.