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

We’re here with Glenn Kowack, who’s chief exec­u­tive of EUnet. Glenn, I’d like to wel­come you to Geek of the Week.

Glenn Kowack: Well thanks for invit­ing me. Happy to be here.

Malamud: Great. EUnet. In the United States we have some­thing called UUNET, which is a large com­mer­cial TCP/IP and UUCP ser­vice provider. Is EUnet the same thing, or is it a dif­fer­ent type of service?

Kowack: Well, first of all we have an E in place of a U. So there’s at least a fun­da­men­tal dif­fer­ence there— No, we are very sim­i­lar in our ori­gins, and in the sorts of ser­vices we have. And in some sense in our geo­graph­i­cal cov­er­age. But to set the cir­cum­stances in Europe in terms of polit­i­cal and geo­graph­i­cal bound­aries are real­ly quite dif­fer­ent from the cir­cum­stances in the States. Briefly put, we are an essen­tial­ly com­mer­cial provider of Internet-related ser­vices in the European region. Quite sim­i­lar to the sorts of things you see in UUNET, and the region­als, in the United States although in terms of geo­graph­i­cal cov­er­age we would be a very large region­al indeed.

Malamud: How many coun­tries do your net­works span?

Kowack: Today we say about twenty-three.

Malamud: And you oper­ate net­works in every one of those countries?

Kowack: Yes we do, although in vary­ing degrees of sophis­ti­ca­tion due to the vary­ing tech­ni­cal cir­cum­stances in Europe. As you might imag­ine, a net­work in the Netherlands is going to be much more sophis­ti­cat­ed than a net­work that’s just start­ing up in for instance Poland or the Ukraine.

Malamud: And you have net­works in coun­tries like Poland and the Ukraine.

Kowack: I picked Poland and the Ukraine because they’re two places—in one case Poland where we’ve been try­ing to estab­lish a net­work and hav­ing some dif­fi­cul­ty. The Ukraine is a place where we have some over­tures going out where we’re try­ing to devel­op those. Both coun­tries have daunt­ing prob­lems because of their local infra­struc­ture, lack of tech­ni­cal sophis­ti­ca­tion broad­ly through­out the pop­u­la­tion, and then just sort of dis­tance from…let’s say Western telecom­mu­ni­ca­tions con­nec­tiv­i­ty and tra­di­tions. We do pride ourselves,though, on hav­ing a lot of con­nec­tiv­i­ty to Eastern Europe. In fact EUnet in the late 80s and ear­ly 90s was one of the first net­works to get into Czechoslovakia, Hungary, the Soviet Union as it was known in those days, and the for­mer Yugoslavia as well.

Malamud: Now, you tend to struc­ture your activ­i­ties around a nation­al Unix users group. Can you tell me a bit about how that gets set up? Let’s say we’re in some new coun­try, Freedonia, [Kowack laughs] and we wish to join the Internet. How does that work?

Kowack: Uh, Freedonia’s an excel­lent choice because the Marx Brothers— Well, nev­er mind. I’ll get into that lat­er. The typ­i­cal cir­cum­stances through which we start a nation­al net­work are… Well, let me do a lit­tle bit of his­to­ry. I hope this isn’t too long. 

EUnet was start­ed around 1982 under the umbrel­la of the European Unix User Group, now known as EurOpen. In 1982, we would have con­fer­ences of peo­ple around Europe from dif­fer­ent insti­tutes, all of whom were Unix enthu­si­asts. These peo­ple were aware of what was going on in the United States with the Usenet and the UUCP net start­ed by peo­ple like Tom Truscott and oth­ers in Research Triangle Park in North Carolina in the States. And these peo­ple got togeth­er at one of these meet­ings and they said well look, we’ve got local data tra­vers­ing our local net­works, we’ve got some nation­al activ­i­ty. But right now there’s no inter­con­nec­tion mod­el inside Europe and we need one. So they very infor­mal­ly agreed to just start con­nect­ing each oth­er. And it took off like wildfire. 

One of the ways that we’re very dis­tinct from US net­work­ing is that in some sense EUnet and most of the rest of Europe start­ed out with region­als and is today real­iz­ing an NSFNET equiv­a­lent. In EUnet’s case we had sort of very loose insti­tu­tion­al nation­al net­works. They start­ed com­ing togeth­er to inter­change traf­fic in the way that I men­tioned a minute ago. And it has blos­somed since then. From rough­ly 1982—a lit­tle bit lat­er, until more or less today, we have oper­at­ed under the umbrel­la of the EurOpen group, which again is the mod­ern man­i­fes­ta­tion of the European Unix User Group. 

In order to make our oper­a­tions work on a nation­al lev­el we have an affil­i­a­tion between a local Unix user group and the net­work­ing group that start­ed out. In essence, we allow our­selves to be start­ed by a user group. That ensures that we are watch­ing out for our users, for their require­ments, and be sen­si­ble about how we pro­vide our ser­vices. So when we want to start a net­work in anoth­er coun­try, typ­i­cal­ly what we do is we ask them to come togeth­er and start a EurOpen group. Now, that usu­al­ly requires some sort of Unix enthu­si­asm but in fact that’s evolved in recent years to enthu­si­asm towards open sys­tems which is near­ly syn­ony­mous with open net­work­ing now. And as that group devel­ops, we work with tech­ni­cal peo­ple asso­ci­at­ed with that group to start some kind of local group. And once that group is iden­ti­fied we imme­di­ate­ly start to exchange traf­fic with them, usu­al­ly by UUCP, which works quite well over the low-grade phone lines that we’ve seen in recent years in Eastern and Central Europe. 

And then as time goes on, as the group becomes more cohe­sive, iden­ti­fies itself as being a group that can deliv­er ser­vices, starts bring­ing in some fund­ing, then slow­ly we move to more sophis­ti­cat­ed ser­vices. IP, pos­si­bly over dial-up, even­tu­al­ly over leased lines. 

Malamud: You have you have nation­al net­works that actu­al­ly have leased lines in place.

Kowack: In fact today we have I believe eleven nation­al net­works that are con­nect­ing to the Amsterdam European Network Operations Center for EUnet with leased lines. And the vast major­i­ty of those are 64 kilo­bits; a few are 128, notably Germany.

Malamud: That’s a fair­ly decent-sized oper­a­tion. These are not cheap line in Europe.

Kowack: No. They’re extreme­ly expen­sive. Probably… What do you think the bench­mark is? Maybe a fac­tor of three times more expen­sive for leased lines in Europe, in your experience?

Malamud: Well that sounds like a good benchmark. 

Kowack: Yeah I would say so. And transat­lantic lines are the same. We have today a 256 kilo­bit line that’s just being installed, upgrad­ing from our 128k line. And it is unfor­tu­nate that European telecom­mu­ni­ca­tions pol­i­cy, dom­i­nat­ed by the PTTs in near­ly every coun­try, has imposed a hid­den tax on its pop­u­la­tion. But now, admit­ted­ly there are com­plex ques­tions regard­ing uni­ver­sal voice phone ser­vice, so that say­ing it’s pure­ly a tax is per­haps a lit­tle bit harsh. But I’ve been now doing this for a few years and feel like a lit­tle harsh­ness is appro­pri­ate. The lines are just so expen­sive and it’s hold­ing back European net­work­ing so much. It’s real­ly sad. 

And then there are terms and con­di­tions that the PTTs impose on one. Like in some countries—Ireland—you need to order a leased line and pay for it one year in advance. In oth­er coun­tries there are dra­con­ian ter­mi­na­tion claus­es, includ­ing the Netherlands. It makes net­work oper­a­tions very dif­fi­cult but then again that pro­vides for a niche for EUnet, because we pride our­selves in find­ing ways to get things done in spite of the impediments.

Malamud: Now, your users pay for their ser­vices. You don’t get any gov­ern­ment subsidies?

Kowack: Well, we nev­er get any sub­si­dies. We do have some grants com­ing in. Some of our peo­ple in for instance Ireland and Denmark are work­ing in the CEC VALUE pro­gram. But those are for spe­cif­ic imple­men­ta­tions of a ser­vice which is then being pur­chased by the pro­vid­ing orga­ni­za­tion. No, we real­ly stand on our own eco­nom­ic feet. We don’t receive any sub­si­dies, we don’t receive any major grants on a con­tin­u­ing basis. We are depen­dent upon oper­at­ing our­selves as a viable eco­nom­ic unit, and frankly I like it that way. 

As Piet Beertema has said— Piet Beertema’s one of the long-term lead­ing lights of EUnet, hav­ing been involved since the ear­ly 80s—he’s our head of oper­a­tions today. And Piet has always done a very excel­lent and insight­ful job of encour­ag­ing peo­ple to nev­er dis­tort their cost eco­nom­ics through grants. It may seem like a con­tra­dic­tion, and it may seem per­haps like an impo­si­tion on the user, but in fact if you’re start­ing a net­work which has a sub­scrip­tion cost which does not reflect the actu­al cost, then at some­time in the future you’re going to have to charge those costs. And that’s a square wave inser­tion of a new price. And that can destroy net­works. And in fact that’s one of the things that has made some of the net­works in Europe have a very very dif­fi­cult time. They get sub­si­dies for a long time and that they watch the cliff com­ing, the cliff of when those sub­si­dies change or end, and…well, they’re not sure they can bridge the gap. It’s a tough situation.

One of the things that our eco­nom­ic mod­el forces EUnet to do, which I think is very inter­est­ing and good for us, is it forces us to smooth out our price and cost curve when­ev­er pos­si­ble. In oth­er words, instead of hav­ing a fetish of 2 megabit infra­struc­tures or 34 megabit infra­struc­tures that you see in a lot of net­work­ing around the world and I think par­tic­u­lar­ly in Europe as in, ​“Let’s build a big net before we have any net at all.” What EUnet has said across the length and breadth of the European region is, ​“Let’s do any net­work­ing we can’t. Let’s bring in mon­ey. And as demand grows, and as funds accu­mu­late, let’s go to the next step.” We have very low over­head. We put most of our mon­ey back into the tech­nol­o­gy. And if our European-level orga­ni­za­tion has any sur­plus­es we grant mon­ey back to the dif­fer­ent nation­al net­works so that they can grow faster and get the net­work­ing going. We think that EUnet is proof that one can suc­ceed in…how should I say, self-built infra­struc­ture. It does­n’t require ter­ri­ble gov­ern­ment inter­ven­tion, and in fact in some ways it does bet­ter with­out it. 

Malamud: You’re lis­ten­ing to Geek of the Week. Support for this pro­gram is pro­vid­ed by O’Reilly and Associates, rec­og­nized world­wide for defin­i­tive books on the Internet, Unix, the X Windows sys­tem, and oth­er tech­ni­cal top­ics. Additional sup­port for Geek of the Week comes from Sun Microsystems. Sun, the net­work is the computer.

Don’t touch that mouse. Internet Talk Radio will be right back. 

[Ask Dr. SNMP seg­ment omitted]

Malamud: Well now, in many coun­tries, there’s two net­works. There is the EUnet affil­i­ate, and there’s a nation­al R&D‑sponsored net­work, which is typ­i­cal­ly sub­si­dized, typ­i­cal­ly aimed at the research and devel­op­ment com­mu­ni­ty. Do you com­pete with those net­works, for customers?

Kowack: Um…sadly some­times we do. But by and large we don’t because of two dis­tinc­tions. The first is that it’s sim­ply impos­si­ble for us to com­pete in any long-term fash­ion with net­works that are heav­i­ly sub­si­dized. Now, like in the US, at least in the past—I’m not sure of the cir­cum­stances today—many of the nation­al research net­works were found­ed by their nation­al research boards. Funding was made avail­able by those nation­al research report to the user orga­ni­za­tions for that net­work. And the fund­ing was ear­marked to go direct­ly to that net­work. Well, then an EUnet rep­re­sen­ta­tive shows up and says, ​“Well, would you like to get a ser­vice from us? It cost so much.” And it sim­ply does­n’t work. 

When the nation­al net­works first start­ed out in some of the coun­tries of Europe… Well, let me do a lit­tle bit of back­ground because it’s quite impor­tant. EUnet start­ed around 1982, as I men­tioned, and was quite siz­able soon there­after. Most of the nation­al research net­works in Europe are phe­nom­e­na of the mid to late 1980s. So in fact EUnet nation­al net­works pre­date the research net­works by quite a few years. So, typ­i­cal­ly what we saw was in coun­try X in Western Europe we had a net­work up and run­ning. It was pro­vid­ing a vari­ety of connections—sometimes IP, most­ly UUCP. And then the nation­al research net­work came. The first few times this hap­pened, as they start­ed to drain away our sub­scribers, there was quite a bit of dis­may in the net­works. But even­tu­al­ly we learned that there was sim­ply no way to retain sub­scribers under such a base, and we moved on to a broad­er sub­scriber base. 

Now, we’re still pri­mar­i­ly in the R&D com­mu­ni­ty. But we are by and large no longer as heav­i­ly in the domain of users at major insti­tutes. We still pro­vide con­nec­tiv­i­ty to a lot of major insti­tutes. We’ve by no means left that area. But it’s just no longer our sin­gu­lar thrust.

Malamud: What types of users do you have, or are you begin­ning to attract?

Kowack: Well, let’s draw the big pic­ture across Europe. We have today about forty-five hun­dred sub­scrib­ing net­works and sites. Of those forty-five hun­dred, prob­a­bly on the order of forty-one hun­dred or so are in fact UUCP sites. Now, some of those UUCP sites can get fair­ly big in some coun­tries. For rea­sons I’m not entire­ly clear on some of our sub­scribers insist on twenty-four-hour UUCP lines, as opposed to IP. Well, they’ll learn and we’ll move them over even­tu­al­ly. So we’ve got on the order of three- to four hun­dred IP sites. 

We con­nect pri­mar­i­ly to orga­ni­za­tions which start­ed out in the com­put­er sci­ence domain, and have since start­ing out there have moved sort of into the bound­aries of that. We con­nect to insti­tutes, com­pa­nies involved in research, major multi­na­tion­al cor­po­ra­tions, medium-sized com­pa­nies, and fun­ni­ly enough not that many one-man shops—or one- or two-man shops. We usu­al­ly are in the large- to medium-size enter­pris­es. I think in the long run it’s not clear who we will be pro­vid­ing ser­vices to, because the mar­ket remains high­ly undif­fer­en­ti­at­ed. There’s a lot of mar­ket learn­ing to go on. 

But our goal is to pro­vide ser­vices to whomev­er needs it, with a few excep­tions. We are not in the high-availability, high-reliability net­work­ing mar­ket­place. If you want a net­work that stays up 99.9999% of the time, we don’t com­mit to that lev­el. And you should find some­one else to do it. At least that’s the next two years or so. In the future we’ll see where that evolves. We do not pro­vide real-time net­work­ing in the sense of a guar­an­teed deliv­ery of pack­ets with­in some nar­row time region. And we’re not pro­vid­ing ser­vices in what we’d call the highly-secure mar­ket­place, which to us would include for instance finan­cial trans­ac­tions. That’s not our busi­ness. Essentially we’re sci­en­tif­ic, tech­ni­cal, and then relat­ed businesses.

Malamud: Do you think that EUnet’s going to be able to com­pete in the long run with com­mer­cial providers like Infonet or like SWIFTnet in Sweden, or with Sprint, who is rolling out a com­mer­cial IP operation?

Kowack: I’d like not refer to any spe­cif­ic com­pa­nies, because any one would take a lengthy discussion—

Malamud: Okay, well Freedonia Networking Incorporated.

Kowack: [laughs] FNI, right. Or FNL for ​“Limited.” Um. I think we do com­pete today and I think we can com­pete, pri­mar­i­ly because of three things. Number one, we are a long par­tic­i­pants in the net­work­ing world and we have excep­tion­al expe­ri­ence in the tech­nol­o­gy. We under­stand the tech­nol­o­gy. Number two, we have a large installed base. And we can use that as a basis for growth. We know how to get to sub­scribers. And then final­ly, we have enthu­si­asts who real­ly know this mar­ket, who’ve been around for a long time, and are good at chang­ing as the envi­ron­ment changes. 

One of the bases of that that is very pow­er­ful is that we’re a light­weight orga­ni­za­tion, in the sense that we do not have mas­sive over­head, we don’t have a large union­ized staff. Most of our oper­a­tional cen­ters are three to six peo­ple in a giv­en coun­try. We have a lot of expe­ri­ence in using the appropriately-priced, appropriately-technical equip­ment to get our job done. We under­stand how to march up the smooth price curve and bring in the sub­scriber base as we go. And I think if we look across the Atlantic River to expe­ri­ence in the United States, some of the net­works that have ori­gins sim­i­lar to ours are quite suc­cess­ful and lead­ing what look like the highly-capitalized big telecom-oriented companies. 

But then again, it is a coin toss. And we’ll see what the future gives us.

Malamud: So you think there’s a role for a not-for-profit orga­ni­za­tion such as yours in a com­mer­cial world. You don’t think com­mer­cial net­work­ing is going to take that over.

Kowack: Um, I think that in the long run there’s prob­a­bly no such thing com­mer­cial net­work pro­vi­sion. There maybe com­mer— I’m sor­ry, there’s no such thing as not-for-profit net­work ser­vice pro­vi­sion in terms of leased lines and things like that. Because the risks are too great. You move too much cap­i­tal, you have too many com­mit­ments. And you sim­ply need to have the pos­si­bil­i­ty of reward for the risks that come into place. It’s not viable sur­round­ed by foun­da­tions. And EUnet is mak­ing that evo­lu­tion as we speak. 

Now, I spoke ear­li­er that EUnet is sort of a com­mer­cial ser­vice provider. And I’d like to actu­al­ly retreat from that because…I’m not sure what com­mer­cial is and what com­mer­cial isn’t, in Europe or the United States for that mat­ter. Networking is such a jumbled-up domain in terms of what’s com­mer­cial, what’s not; what sort of traf­fic is allowed, what’s not. Are we in a pre-competitive phase, are we in a post-competitive phase? We seem to be the philoso­pher’s ele­phant that’s described by so many blind philoso­phers at the same. 

What EUnet does is to pro­vide ser­vices, broad­ly to peo­ple who need those ser­vices, and we accept sub­scrip­tion funds for those ser­vices. We work in some coun­tries in a not-for-profit fash­ion. In oth­er coun­tries, notably let’s say Ireland, we have a for-profit com­pa­ny set up—or Switzerland—simply because in the case of Switzerland there’s no such thing as a not-for-profit company. 

What we’ve said his­tor­i­cal­ly is that we are fre­quent­ly a for-profit struc­ture with not-for-profit goals. But that’s evolv­ing. Technical enthu­si­asm with new tech­ni­cal ideas will wane as the ideas become mature. And things like IP tech­nol­o­gy are no longer the spank­ing new, brand new excit­ing infants that they once were. We’re learn­ing more about it. So we will become in time more com­mer­cial, and more for-profit in our structure.

Now, in the last two years we’ve seen an evo­lu­tion of our core of about twelve net­works in west­ern Europe from most­ly being infor­mal asso­ci­a­tions to pri­mar­i­ly being com­mer­cial com­pa­nies in their for­mal struc­ture. They’re still large­ly work­ing with a not-for-profit struc­ture but that’s chang­ing very very fast and I sus­pect that some of your lis­ten­ers who’ll be lis­ten­ing to this in the mid­dle and lat­er in 1992 may have have also heard about more announce­ment and changes in that direction.

Malamud: There are many types of net­work providers in Europe. Many of them come out of the European Commission-sponsored activ­i­ties, such as the COSINE project, which has an X.25 base. EUnet seems to have been able to avoid the technical-religious impuls­es of using X.25 and OSI. Does that cause you prob­lems in Europe?

Kowack: [laughs] Well it has caused more prob­lems in the past than it caus­es now. EUnet has because of its eco­nom­ic base or lack there of in terms of sub­si­dies, has had no choice but to do what works. And to do what peo­ple will pay for. I’ll say it again, we’ve had no choice. So being an agent for you European indus­tri­al pol­i­cy, or EC indus­tri­al pol­i­cy, is sim­ply not avail­able to us. We can’t do it. So when it came time to deliv­er a ser­vice and we tried at some point or anoth­er to deliv­er X.25 or or OSI or what have you, it sim­ply did­n’t have the same kind of pop­u­lar response as you UUCP, RFC 822 email, and so on and so forth. And…simply put in some sense our users have decided. 

I like to say that we are the Deng Xiaoping of European net­work­ing. We don’t care what col­or the pack­ets are as long as they catch data. And it has caused us some con­flicts in years past. We in fact had a migra­tion plan to go over to OSI cir­ca 1989, but it…it sim­ply fad­ed. It was­n’t the appro­pri­ate thing to do.

Now, what’s been so inter­est­ing in the last two and a half years is that clear­ly IP has won the day in terms of prac­ti­cal net­work­ing. All the major net­work­ing in Europe today, and IXI [pro­nounced ​“ick­see”] pro­po­nents, inde­pen­dent of what they claim—

Malamud: IXI is the X.25 infra­struc­ture [crosstalk] spon­sored by the commission.

Kowack: Right. Exactly. So the inter­im X.25 infra­struc­ture I think is what it was called. IXI nev­er got, dur­ing its projected—or its originally-planned life­time over 64k in terms of its leased line capac­i­ty. The per­for­mance I think was lack­ing. Its pop­u­lar­i­ty was also lack­ing. And in the mean­time the IP net­works around Europe explod­ed. Proponents in nation­al net­works of X.25 and OSI net­work­ing have qui­et­ly in the last few years become users, the best advo­ca­cy of all, of IP networking. 

Now, in the long run who knows what we’re going to see. If you’d asked me three or four years ago if OSI—or even two years ago—if OSI was going to win the day I’d say def­i­nite­ly; it was just a mat­ter of time. Now I don’t know. I read things by Marshall Rose and I watch sort of the tone of the net­work­ing at large, and it’s a coin toss. But I’m very impressed at how the Internet stan­dards seem to be win­ning the day in terms of going toward the future the best and most quickly.

But quite hon­est­ly, to back up again, we don’t view our­selves as net­work tech­nol­o­gy big­ots. If OSI turns out to be the right imple­men­ta­tion, with the right per­for­mance, with real prod­ucts avail­able, and manageable—notice I made all those qualifications—

Malamud: You believe in miracles.

Kowack: [laughs loud­ly] Well I’m will­ing to accept them, but the belief is a lit­tle bit of a stretch. If all those things happen—and they’re not here today—then let’s use OSI. I don’t care. Why should I? I mean, what do net­work users want? How do we advance that? That’s real­ly the question.

Malamud: You’re lis­ten­ing to Geek of the Week. Support for this pro­gram is pro­vid­ed by Sun Microsystems. Sun Microsystems, open sys­tems for open minds. Additional sup­port for Geek of the Week comes from O’Reilly and Associates, pub­lish­ers of books that help peo­ple get more out of computers.

This is Internet Talk Radio. You may copy these files and change the encod­ing for­mat, but may not alter the con­tent or resell the pro­grams. You can send us mail to mail@​radio.​com.

Internet Talk Radio, same-day ser­vice in a nanosec­ond world.

Malamud: A prac­ti­cal bent is evi­dent in the fact that EUnet is often the very first net­work­ing in a coun­try. Can you tell me what that’s like, [Kowack laughs] show­ing up in Algeria or Tunisia or some coun­try and start­ing from scratch? What do you do? What do you do to get things moving?

Kowack: Um… Well, let me give you an exam­ple in Bulgaria. We were work­ing per­haps a year and a half ago, two years ago, at try­ing to find good con­tacts in the new­ly opened-up Eastern European area. And we start­ed get­ting some mail from Danbo, this… Excuse me, I think it was a… It migh­ta been a FidoNet site at the time. And we start­ed get­ting some email. And Piet Beertema whom I men­tioned before, our head of oper­a­tions, was doing a fair amount of con­ver­sa­tion across email with this Danbo guy or who­ev­er it was. And after a month or two, Piet said, ​“Well look, we’ve got this guy Danbo out here in Bulgaria. He seems to be real­ly bright. Why don’t we just let him exchange email with us.”

So, we had­n’t had any such thing quite before this but we declared him a coop­er­at­ing site. What that means is we’ll exchange email with him. We don’t care who he is… Oh, as long as it’s not mil­i­tary. We don’t have a mil­i­tary traf­fic. We want to stay com­plete­ly out­side of that domain.—

Malamud: And did you have any evi­dence this was­n’t the Bulgarian secret police?

Kowack: Um. Well, we don’t have any evi­dence that any­one’s not any­thing in par­tic­u­lar. All we knew is was that he was­n’t going to exchange any data at a ter­ri­fy­ing­ly fast rate. And if he was the secret police he’d prob­a­bly have some expen­sive equip­ment which he obvi­ous­ly did­n’t have.

So, we start exchang­ing data with him. We invit­ed him to one of our next meet­ings. We sent Norman Hall, EurOpen’s start­up guy, from Ireland to fly to Bulgaria and have a meet­ing, which he did. We got a user group start­ed straight away. And this wild­man from Bulgaria who’s an absolute­ly bril­liant net­work­er by the name of Daniel Kalchev shows up at one of our meet­ings with just amaz­ing ques­tions and ideas and, ​“Where do I put my IP line?” and so on and so forth. In oth­er words he was way ahead of us. 

And so we now have an offi­cial site in Bulgaria. We exchange data with Digital Systems, which is this group that oper­ates the net­work there. And it works quite well. We start­ed off by doing UUCP mail. I think today he’s exchang­ing data via X.25. Oh and by the way, EUnet is a fre­quent user of X.25. We were at one point a sub­stan­tial user of IXI. But we have moved on to oth­er things since then. So, we’re using X.25 in Bulgaria. And it’s work­ing quite well. 

We’ve had sim­i­lar expe­ri­ences in Hungary and in Yugoslavia, Czechoslovakia. Typically what we do unlike Bulgaria’s we start out exchang­ing mail with some insti­tute. And then after a year or two or three as the mail exchange group in the insti­tute grows to be big­ger than the insti­tute (which has hap­pened in some cas­es) well then they have to spin off and start an orga­ni­za­tion of some kind. And they fol­low what­ev­er’s the struc­ture in the in the coun­try in question. 

Usually what hap­pens is a starting-up net­work con­nects either direct­ly to Amsterdam via dial-up. Again, Amsterdam is where we have our European NOC. Or they con­nect via some inter­ven­ing EUnet group. Now, the phone leased line costs in Europe are crazy enough so that almost essen­tial­ly in Europe, once you cross the bor­der it does­n’t mat­ter how many bor­ders you cross in Europe. It just becomes a pan-European con­nec­tion. So fre­quent­ly it makes a lot of sense for a group let’s say in…Country X to con­nect to Amsterdam even though there may be inter­ven­ing nets in between. It does­n’t save them any mon­ey, and it would just have inter­ven­ing hops that aren’t useful. 

But I have ranged a bit far afield [crosstalk] on the start­up question.

Malamud: No no, that’s fine. I under­stand that you actu­al­ly once had a real job. You man­aged R&D for a large com­put­er com­pa­ny. You worked on secure Unix imple­men­ta­tions. How did you ever get into this posi­tion where you’re going to places like Bulgaria and set­ting up nation­al networks?

Kowack: Um… Well. [laughs] Um…in 1989 I was work­ing as a con­sul­tant for among oth­er peo­ple Unix International and Harris Semiconductor. And I had been for years fol­low­ing the polit­i­cal sit­u­a­tion in Eastern Europe but par­tic­u­lar­ly the works of Andrew Weschler [pos­si­bly Lawrence?] in The New Yorker mag­a­zine as he report­ed on the sit­u­a­tion with Solidarity. And I was enam­ored of the sophis­ti­ca­tion, the polit­i­cal sophis­ti­ca­tion of this group that was try­ing to cre­ate a new free­dom in their country. 

And in ​’89 of course things start­ed to speed up quite a bit. And I was watch­ing very close­ly while doing all this con­sult­ing work in Florida and New York. And at one point I was stand­ing with my friend David Eisenman[sp?] in his kitchen dur­ing the win­ter, and it was I think December 11^th or 13^th, 1989, and the Berlin Wall had just fall­en. And the Velvet Revolution I think was either in mid-stream or in start. And Poland had opened up tremen­dous­ly with the Solidarity/government talks start­ing in April and lat­er of ​’89. And Eisenman turns me and he goes, ​“Kowack, you are uen­cum­bered, you’re unat­tached. You should go to Poland. You should…you know, help them you know, inte­grate with the West. Start a com­put­er com­pa­ny. You can do anything.”

“Nah. Impossible,” said I. And a half hour lat­er we had worked togeth­er a plan so I decid­ed to go to Poland. And I spent most of 1990 in Poland try­ing to look for oppor­tu­ni­ties to start a soft­ware devel­op­ment com­pa­ny. That’s a lot of the back­ground. I come from [indis­tinct]. I did a lot of soft­ware devel­op­ment of net­work­ing tools. And while in Poland, Czechoslovakia, and Hungary—I spent about maybe six or eight months and in those coun­tries in total—I did what I could to help start up or help Unix groups in those areas, includ­ing by using their email a lot. It turned out to be one of the most reli­able ways I could get to friends in the States, phone con­nec­tiv­i­ty and prices being so prohibitive. 

By the way, as an aside. Just to give you a sense of what it was like to do some­thing in Poland in 1990. I think…at least I was told that in 1990 around June, there were six­teen direct lines from Poland to the United States. Now, this direct phone lines…uh, let’s think about this for a sec­ond. Chicago is the second-largest Polish city in the world, right. And there are six­teen lines between Poland and the United States. I mean, it’s unbelievable. 

As an aside that’ll show you how hard it is to do net­work­ing in some of these coun­tries, the pol­i­cy of the PTT in Poland, or the tele­phone com­pa­ny in Poland, was two-fold and very sim­ple. Number one, have a work­ing tele­phone sys­tem. And num­ber two, don’t allow peo­ple to com­mu­ni­cate. This was the pre-1989 rule. And as you can imag­ine it cre­at­ed all sorts of inter­est­ing prob­lems. For instance, the sys­tem of exchanges in Warsaw was not ful­ly inter­con­nect­ed. There were no islands that I know of, but very often, one exchange could not talk to anoth­er exchange across the street. So, at least in the mid 80s, accord­ing to Polish friends, the way that you got some­body togeth­er for a meet­ing is you sent them a telegram. It’s not clear how much of this was by pol­i­cy or by acci­dent or by eco­nom­ic incom­pe­tence. Maybe there were just bro­ken lines, but it was like this all the time and at ran­dom you would­n’t be able to call between dif­fer­ent parts of town. 

Which leads me to a won­der­ful anec­dote about net­work­ing. When I was in Poland I helped to start at least the begin­nings of a Polish Unix user group which has floun­dered since then, I’m sor­ry to say. But while there I vis­it­ed a lot of com­pa­nies look­ing at the pos­si­bil­i­ties of help­ing them get con­nect­ed to EUnet, and start­ing some kind of soft­ware ven­ture. So here I am in this com­pa­ny just on the out­skirts of Warsaw, which is um…very…very California-like. A lot of peo­ple there are mov­ing real fast; it does­n’t feel like a reg­u­lar Polish office. The aver­age Polish office felt like a dete­ri­o­rat­ed ver­sion of a Mission Impossible set at the time. But here I’ve got this house on the out­skirts of Warsaw, it’s real­ly neat, a lot of fast-moving people. 

And I’m sit­ting down with these guys explain­ing how they can use EUnet. Remember, I don’t work for EUnet, I’m just wan­der­ing using EUnet and telling peo­ple about it. And I’ve got their head tech­nol­o­gist, who looks like Rasputin; he’s got long brown hair and he’s obvi­ous­ly been up for three days in a row pro­gram­ming some­thing on a PC. And I say, ​“It’s real sim­ple. All you need to do to get con­nect­ed to EUnet is get some free UUCP soft­ware. It runs on PCs. Absolutely no prob­lem. You’ve got modems here. Just dial up to a con­cen­tra­tion point in Warsaw which we’ll bring in. And then you can get and exchange your mail with the West to bring up a cus­tomer base, do busi­ness around the world, and so on and so forth.”

And he goes, ​“It’s impossible.”

I said, ​“Why is it impossible?”

He says, ​“We’ve only got one modem.”

I said, ​“You’ve only got one modem. Well, we’ll get you anoth­er modem.”

He says, ​“It’s impossible.”

I said, ​“Why is it impossible?”

He says, ​“Well—” By the way, this is ver­ba­tim true. I could not have made this up. It’s such a strange event. He says it’s still impossible.

“Why.”

He says, ​“Well. We’ve only got one phone line.”

“You’ve only got one phone line. There are forty peo­ple in this company.”

“Yeah, we’ve only got one phone line.”

I said, ​“Well, what’s the prob­lem with that?”

He says, ​“Well, we have to use it.”

“Well, use it at two or three o’clock in the morning.”

He goes, ​“I can’t come in every day at two or three o’clock in the morning.”

I said, ​“Well, god. You know, put a cal­en­dar pro­gram in your PC. Have it wake up auto­mat­i­cal­ly. Have it do what­ev­er you need.”

He says, ​“No no no. It’ll both­er the peo­ple next door.”

“Why is it gonna both­er the peo­ple next door? I mean, how loud can a modem be?”

He sayd, ​“No no, it’s their phone line. There’s a res­i­den­tial house next door.”

I go, ​“What do you mean?”

He said, ​“It’s a par­ty line. You got­ta come in and push the but­ton to switch the line over to our line from their line and they use it at night.”

I said, ​“Well—” I said, ​“Okay. So you’re say­ing some­body has to come in at night to press the button.”

“Yes.”

I said, ​“Well you’ve got forty peo­ple here. Just have… You know, every­body’s on a rotary. So one per­son comes in a dif­fer­ent day.”

He says it’s impos­si­ble, it won’t work. I said, ​“Why?”

He goes, ​“The lines are too noisy.”

And we went on like this for about twen­ty min­utes. It was one of the most aston­ish­ing things that I’d ever seen. And he was right about every­thing. But the thing that he was­n’t right about, and I think he’s dis­cov­ered that since then, is that in fact things did change and you could make progress. And you could move things forward.

Malamud: So it was­n’t impos­si­ble. He was actu­al­ly able to even­tu­al­ly get his con­nec­tions up.

Kowack: I think so. Yeah. I don’t have a hard con­fir­ma­tion on it but as far as I know he did. Just because in com­mu­ni­ca­tion with friends in Poland and oth­er Eastern European coun­tries the telecom­mu­ni­ca­tions aren’t won­der­ful but they’ve improved fast. And you can do a lot of things. 

So it was just a fas­ci­nat­ing expe­ri­ence of talk­ing to some­one who only saw imped­i­ments and did­n’t see solu­tions. Now, he deserves spe­cial con­sid­er­a­tion because in fact the cir­cum­stances were unique­ly dif­fi­cult there. It seemed like every time you found a solu­tion some­one would find a wall to put in front of you. But in fact the EUnet mod­el, and the mod­el of a lot of net­work­ers around Europe not just EUnet, is just find a way to get to the next step. And that’s one of the real­ly nice things about net­work­ing in gen­er­al. It’s full of a lot of peo­ple who are ener­getic, cre­ative, and just aren’t real­ly good at spelling the word ​“no.” They just keep chug­ging forward.

Malamud: Or ​“nyet,” as it.

Kowack: Nyet, or nei, or nein, or what­ev­er you’ve got this week, right.

Malamud: This has been Geek of the Week, brought to you by Sun Microsystems and by O’Reilly and Associates. To pur­chase an audio cas­sette or audio CD of this pro­gram send elec­tron­ic mail to radio@​ora.​com. Internet Talk Radio, the medi­um is the message.