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

This is Carl Malamud. We’re in Amsterdam with Rob Blokzijl, who is net­work man­ag­er at NIKHEF, that’s the Dutch High Energy Physics Institute. He’s also the chair­man of RIPE, which is the Reseaux IP Europeens, which is an IP users group for Europe. Rob, wel­come to Geek of the Week.

Rob Blokzijl: Hello Carl.

Malamud: Maybe you could tell us a lit­tle bit about what RIPE is, and how this group got founded.

Blokzijl: RIPE is the group of IP ser­vice providers in Europe. That is how it start­ed. Nowadays we also have large user com­mu­ni­ties who par­tic­i­pate in the work. The aim of RIPE is two-fold. It is edu­cate peo­ple in the use of large-scale IP net­work­ing, and to coor­di­nate among the ser­vice providers these things that have to be coor­di­nat­ed, like rout­ing, nam­ing, some aspect of appli­ca­tion ser­vices like elec­tron­ic mail.

Malamud: So I guess the first ques­tion that comes to mind then is how does a group like RIPE dif­fer from let’s say the IETF. Do you make standards?

Blokzijl: We don’t make stan­dards. It’s in our char­ter that we do not make stan­dards. People inter­est­ed in net­work research in the sense of devel­op­ing new stan­dards, writ­ing stan­dards, they are strong­ly advised to active­ly par­tic­i­pate in the rel­e­vant IETF work­ing groups.

Malamud: Now why would you put that in your char­ter, that you don’t make stan­dards? Is there some polit­i­cal rea­son for that?

Blokzijl: I think that too many stan­dards already, standard-making bod­ies. And if you have decid­ed to build TCP/IP net­works, there is absolute­ly no need for a European IP stan­dard. That would be extreme­ly dan­ger­ous, in our view. 

Malamud: So you don’t [inaudi­ble] to pro­file the standards.

Blokzijl: Absolutely not.

Malamud: Congratulations. [Blokzijl chuck­les] That of course brings up the ques­tion of why isn’t there a ROSIE or some oth­er group, a Reseaux OSI Europeens. Is there a sim­i­lar group of OSI users out there?

Blokzijl: Uh… I think that’s a…a com­pli­cat­ed ques­tion. OSI is there, but are there OSI users of any large-scale net­works, I mean.

Malamud: But why is IP specif­i­cal­ly in your name? Is there a rea­son that you lim­it­ed your­self to IP net­works and did­n’t draw a broad­er gamut for net­work­ing users.

Blokzijl: Yes. The European fla­vor of net­work­ing dri­ven by pol­i­tics and politi­cians is OSI net­work­ing. At least that was the case a cou­ple of years ago. And all the offi­cial net­work­ing orga­ni­za­tions…RARE is a good exam­ple of such one, they all adopt­ed OSI as the only true reli­gion and were thus not inter­est­ed in real net­work­ing. Real net­work­ing defined as what users can use—today, or three, four years ago.

So, RARE as a European orga­ni­za­tion has nev­er been involved in run­ning EARN BITNET ser­vices, or run­ning DECNet ser­vices, or run­ning IP ser­vices. And just as real users with real needs have found it, EARN have found­ed an orga­ni­za­tion that coor­di­nates DECNet net­work­ing on an inter­na­tion­al scale in Europe. So, when the inter­est for IP net­work­ing more or less explod­ed in Europe three, four years ago, RIPE was founded. 

Malamud: RIPE has recent­ly opened a net­work con­trol cen­ter, [crosstalk] an NCC. 

Blokzijl: No. Network Coordination Center.

Malamud: Ah. Okay.

Blokzijl: We do not con­trol [both chuck­le] anything.

Malamud: And what do you coor­di­nate, then?

Blokzijl: RIPE start­ed work­ing with a bunch of very enthu­si­as­tic vol­un­teers. And after about two years, the amount of work became too much to be car­ried by vol­un­teers in their spare time. So, among the orga­ni­za­tions involved in RIPE, we decid­ed it’s time to have a more pro­fes­sion­al, very small core of peo­ple who are respon­si­ble for the day-to-day coor­di­na­tion of these services. 

Well, what is there to be coor­di­nat­ed? What is the RIPE NCC doing? In the first place, it is the region­al European reg­istry for IP net­work num­bers. European orga­ni­za­tions, small or large, who need an IP net­work num­ber go to the RIPE NCC. This is espe­cial­ly impor­tant with the cur­rent prob­lems of address­ing and rout­ing in the IP world, where cen­tral coor­di­na­tion of the hand­ing out of blocks of num­bers is need­ed in order to be able to do aggre­gat­ed rout­ing in the near future.

Malamud: How did RIPE get to be the NIC. In the US, there’s Department of Defense routes to the NIC and there’s a National Science Foundation award­ing a NIC con­tract. How did RIPE get that func­tion for Europe?

Blokzijl: Uh, for many many years the NIC, wher­ev­er it was locat­ed in the US, has always han­dled European request as well. That was fine five years ago or sev­en years ago when there were not that many European IP net­works, and the few there were were main­ly in the aca­d­e­m­ic and research world—computer sci­en­tists who knew how to han­dle the NIC. But with the emer­gence of more and more non-computer sci­ence com­mu­ni­ties into this field of net­work­ing, it became more and more dif­fi­cult to com­mu­ni­cate with the NIC. There all kinds of sim­ple bar­ri­ers that exist in prac­tice. There’s the lan­guage bar­ri­er; not every­body speaks flu­ent American English. There is the time dif­fer­ence bar­ri­er; there are not many office hours that over­lap between the two con­ti­nents. And there is the knowl­edge bar­ri­er; today peo­ple who apply for an IP num­ber in Europe are still in many cas­es build­ing their first IP net­work. So, they need edu­ca­tion, they need infor­ma­tion from the side of the NIC as well because that’s their first point of infor­ma­tion for the network-building sim­ple exam­ples. Not every­body who is in office imme­di­ate­ly under­stands the fine dif­fer­ences between a bunch of Class C net­works or one Class B net­work, or one Class A net­work. And so out of igno­rance peo­ple may apply for the wrong set of net­work num­bers, and they need edu­ca­tion from a local cen­ter and not from a cen­ter far away on anoth­er con­ti­nent where…and that is only natural…the local knowl­edge is not present.

Malamud: Co you think Europe will then take this to the next log­i­cal step and begin to have nation­al NICs? Will RIPE play a region­al role there?

b We already have many many sub-delegated region­al NICs. Most of the IP ser­vice providers that can be region­al net­works or nation­al aca­d­e­m­ic net­works or com­mer­cial net­works have their own NICs. And these NICs work very close­ly togeth­er with the RIPE NCC.

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 & Associates, pub­lish­ers of books that help peo­ple get more out of computers.

Malamud: You men­tioned that RIPE plays a role in rout­ing coor­di­na­tion, and you’ve men­tioned the IP address space exhaus­tion as issues that you’re begin­ning to tack­le. Could you touch a bit more on the role that you’re play­ing in those areas?

Blokzijl: When we start­ed RIPE, the IP net­work­ing sit­u­a­tion in Europe was very sim­ple. Everybody built his own net­works. We have cho­sen to call all net­works region­al net­works because that leaves out all polit­i­cal dis­pute of who is the only [cul­tured?] nation­al net­work. That was a very excit­ing time. Everybody built his net­work and con­nect­ed it one way or the oth­er to his clos­est neigh­bor net­works, and with­in six months a total chaos was cre­at­ed. So, the need for a rout­ing coor­di­na­tion was very clear to all play­ers. Either play togeth­er, or don’t play at all on the European scale. 

So, one of the first tasks we tack­led was to get some order in the rout­ing coor­di­na­tion. The only way to be found to do that is to have some min­i­mal rout­ing infor­ma­tion stored in a data­base that is being trust­ed by every­body who plays this game. And one of the impor­tant tasks of the RIPE NCC nowa­days is the main­te­nance of this data­base. This data­base which is usu­al­ly approached by the nor­mal whois-type of pro­to­cols, this data­base is now trust­ed by all net­work providers in Europe, and they store all infor­ma­tion need­ed by their col­leagues in Europe to do prop­er routing.

Malamud: So you’re actu­al­ly using the same data­base that’s used for whois as the basis for the rout­ing tables in Europe? 

Blokzijl: Yes.

Malamud: You gen­er­ate the rout­ing tables out of the whois database.

Blokzijl: Yes. The whois data­base, as main­tained by RIPE, has some extra objects and fields that make it pos­si­ble to store rout­ing infor­ma­tion there. What is stored actu­al­ly is not routes them­selves but enough of a pol­i­cy descrip­tion that makes it pos­si­ble for indi­vid­ual IP ser­vice providers to gen­er­ate rout­ing tables from it.

Malamud: I sup­pose hav­ing the whois data­base form the basis for rout­ing infor­ma­tion, that gives peo­ple a very strong incen­tive to keep their whois infor­ma­tion up to date.

Blokzijl: Yes. If you apply this in a few key points on the net­work, it is a great incen­tive to have your infor­ma­tion stored in the data­base. And this was agreed on one of the very first meet­ings of RIPE, and it was recon­firmed lat­er when many more net­works start­ed play­ing this game. And every­body in Europe is deeply con­vinced that this is the only prac­ti­cal way to keep the net­works going. 

One should real­ize that as opposed to the United States, for instance, that there are no…or prac­ti­cal­ly no notions of European-wide back­bones that are cen­tral­ly man­aged. The new EBONE exer­cise that start­ed in ​’92 is the first instance of a centrally-managed backbone-type of net­work in Europe. 

Malamud: Now, EBONe pro­vides a vari­ety of coor­di­nat­ing func­tions for oper­a­tions. There’s anoth­er group called the CCIRN, the Coordinating Council for Intercontinental Research Networking, of which there’s a Euro-CCIRN. Does the role of RIPE con­flict with these groups? Do you find your­self with dif­fer­ent aims, let’s say, than the CCIRN or the EBONE folks, or are you part of the same community?

Blokzijl: It’s the same com­mu­ni­ty. Let’s first talk about the EBONE. EBONE is an impor­tant part of IP net­work­ing today in Europe, but it’s not the only part. So, at a very ear­ly stage when the con­cept of EBONE was born, this has been dis­cussed and every­body con­cerned in Europe agreed that it would be extreme­ly coun­ter­pro­duc­tive to have dif­fer­ent IP groups work­ing doing European-wide coor­di­na­tion, whether it’s a gen­er­al group as RIPE, or a group like EBONE focused on one par­tic­u­lar back­bone net­work. We can’t have two sep­a­rate groups in Europe.

So, all coor­di­na­tion is still being done at the end of the day via RIPE. Of course EBONE…it’s a real net­work. It’s an impor­tant net­work. So they have their own inter­nal tech­ni­cal com­mit­tees to look after their net­work. But at the end of the day it’s a back­bone net­work con­nect­ing to an ever-growing list of region­al net­works, and there is the same need for over­all coor­di­na­tion as that was three years ago when there was no EBONE.

Malamud: Do you see—

Blokzijl: This is— If I may fin­ish. This sounds com­pli­cat­ed, but in the good tra­di­tion of RIPE we keep things very sim­ple. The RIPE rout­ing work­ing group and the EBONE tech­ni­cal com­mit­tee that looks after rout­ing over the EBONE has an over­lap of the 95%, I guess. It’s the same people. 

Malamud: Is there more of a need for coor­di­na­tion in Europe than let’s say in the United States? Is the sit­u­a­tion dif­fer­ent here, and does that mean there’s dif­fer­ent groups that help keep the net­works togeth­er? Is Europe dif­fer­ent somehow.

Blokzijl: Uh… Yes, Europe is dif­fer­ent. United States is one coun­try, Europe is… I don’t know today how many coun­tries. It changes— [laughs]

Malamud: On a dai­ly basis, yes?

Blokzijl: So, we have a lot of country-based, national-based sen­si­tiv­i­ties that we have care of. On the oth­er hand some­times I say okay, we have all these coun­tries in Europe. The United States has a lot of fed­er­al agen­cies doing their net­work­ing, and they need coor­di­na­tion there. It’s a dif­fer­ent lev­el, but the same prin­ci­ple. So yes, there are sim­i­lar­i­ties and there are differences.

Malamud: Do you think region­al coor­di­nat­ing func­tions like you pro­vide for RIPE in Europe is some­thing that we should be see­ing in the rest of the world? Should there be an Asian coor­di­na­tion, should there be a North American coor­di­na­tion group?

Blokzijl: I think yes. I think it’s only nat­ur­al that… If you look at recent devel­op­ments on the North American con­ti­nent with the soon-to-be a free trade zone between the three major coun­tries there, I think it’s only nat­ur­al that this will expand to clos­er coop­er­a­tion in oth­er fields as well, includ­ing sci­ence, research, and com­merce. So there will be a grow­ing need for better-integrated net­work­ing. Of course it’s eas­i­er to coor­di­nate that between three play­ers than between twenty-three play­ers in Europe based on nationalities. 

There is one extra com­pli­ca­tion, I think, one extra dimen­sion in Europe, and that is I have the feel­ing that there’s much more pol­i­tics in net­work­ing in Europe than in the United States. Some of our European gov­ern­ments are not always extreme­ly help­ful in the devel­op­ment of inter­na­tion­al or nation­al networking. 

Malamud: Why is that? And how is that?

Blokzijl: How is that. I think that this is a nice sub­ject for a long series of pro­grams, even. [both laugh]

To give my per­son­al in a nut­shell feel­ing of how it is and why it is, there’s the European Commission in Brussels that decid­ed ten years ago that net­work­ing might be a strate­gic sub­ject for future devel­op­ment in Europe, and they took net­work­ing as one of the fields where they would try to apply indus­try pol­i­cy. And then their choice was very sim­ple: as long as it is not invent­ed in the United States or Japan, it’s all right. And that’s how they became OSI adherents.

And this then trick­led down to the nation­al gov­ern­ments who adopt poli­cies that should at least not con­flict with the one in Brussels. And that’s why we have seen in the past ten years on the nation­al pol­i­cy­mak­ing lev­el a strong lean­ing towards OSI-flavored net­work­ing. And it is some­times amaz­ing to see how long this con­tin­ues, where­as the user com­mu­ni­ty has tak­en quite anoth­er road many years ago already.

Malamud: Is it just stub­born­ness? Is it peo­ple don’t under­stand what’s hap­pen­ing, or they’re unwill­ing, or they think they can still change the tide? I mean, why is there such a strong adher­ence to OSI and X.25-based work­ing and things of that sort?

Blokzijl: I think in the first place the peo­ple who defined this pol­i­cy or are in charge of exe­cut­ing these poli­cies are in most cas­es total­ly igno­rant about what net­work­ing real­ly is. They don’t have ter­mi­nals on their desks. They have not the slight­est idea what they’re talk­ing about. And that’s how once a pol­i­cy doc­u­ment has been signed by twenty-three min­is­ters from twenty-three coun­tries, it’s not easy to change that. If you are a high-level bureau­crat, you keep things like they are.

Malamud: Is there any hope for such an indus­tri­al pol­i­cy? Let’s say there had been a dif­fer­ent group of peo­ple there. Could they have helped, or do you just see that as a lost cause?

Blokzijl: I think it’s a lost cause as far as indus­tri­al pol­i­cy is con­cerned. There is no European com­put­er indus­try. There are no…practically no European man­u­fac­tur­ers of net­work­ing equipment.

Malamud: On the oth­er hand there’s a great deal of tal­ent in Europe, cer­tain­ly in the field of net­work­ing. You know, there’s peo­ple all over the con­ti­nent [crosstalk] who come to—

Blokzijl: Sure. Sure. I mean, all the knowl­edge is here, the expe­ri­ence is here. And so it could be here but I think it’s a lit­tle bit out­side the scope of net­work users to decide that there must be a European com­put­er indus­try. It’s a fail­ure of the com­mis­sion and nation­al gov­ern­ments that the European com­put­er indus­try has disappeared. 

Malamud: This is Geek of the Week, fea­tur­ing inter­views with promi­nent mem­bers of the tech­ni­cal com­mu­ni­ty. Geek of the Week is brought to you by O’Reilly & Associates, and by Sun Microsystems. 

Malamud: We’re begin­ning to see some­thing which I find quite inter­est­ing, which is com­pet­ing ser­vice providers in Europe for net­work­ing. Is that going to con­tin­ue or do you think we’re gonna end up with a sin­gle PTT-like Internet provider for each country?

Blokzijl: Again, it will…for many many years to come it will be dif­fer­ent in var­i­ous coun­tries. But the gen­er­al direc­tion will be the same. It will be towards dereg­u­la­tion. And that means that as soon as there’s a busi­ness case, com­pet­ing net­work ser­vice providers will estab­lish them­selves, either on a nation­al scale or on the European scale. Or on a region­al scale. 

Malamud: To change the sub­ject just a lit­tle bit, one thing I’ve always found very inter­est­ing, if you look at net­work­ing in Europe, it’s like all roads lead to Amsterdam. In fact all roads lead to this build­ing we’re in, which hous­es NIKHEF, it hous­es CWI, which is the Dutch Institute of Higher Mathematics. Why is this small com­plex of build­ings so respon­si­ble for reach­ing out, so will­ing and aggres­sive to bring in net­works from oth­er coun­tries? Is this part of the NIKHEF mission?

Blokzijl: No, absolute­ly not. Our mis­sion is to do par­ti­cle physics.

Malamud: Mm hm.

Blokzijl: And par­ti­cle physics is an expen­sive pro­fes­sion, or hob­by. So in Europe there are only two facil­i­ties where our physi­cists can do their exper­i­ments, and one hap­pens to be in Geneva in Switzerland, and the oth­er in Hamburg in DESY. So, our physi­cists need the best net­work­ing our mon­ey can buy. And that’s why we have been one of the fore­run­ners of inter­na­tion­al net­work­ing in the Netherlands. 

It so hap­pened that when com­put­er net­works start­ed in Europe, one of the real ini­tia­tors was the world of com­put­er sci­en­tists. And that’s how the European Unix Network, which today is a fun­ny name, came about. And it so hap­pened that their net­work is a star, and the star is locat­ed in Amsterdam. So that’s now two net­works on one site.

The com­bined com­put­ing cen­ters of the two Amsterdam uni­ver­si­ties, the com­bined com­put­ing cen­ter, is also on this same cam­pus, and it hous­es some nation­al com­put­ing facil­i­ties like super­com­put­ers. And that’s why the Dutch Academic Network has evolved more or less in into a star cen­tered around, again, the same campus.

Traditionally, the Netherlands have always been an open coun­try in the sense that we are so small and it is such a large world out there. Our econ­o­my is based on inter­na­tion­al trade, inter­na­tion­al ser­vices. So our aca­d­e­m­ic world, the uni­ver­si­ties and high schools, have always been quite open to the world. And if you look at today’s net­work traf­fic on an inter­na­tion­al scale in Europe, you’ll find that one of the small­est coun­tries, the Netherlands, is some­where on top of the list of vol­ume exchange of traf­fic with the NSFNET.

And so, I think the real answer is we have always been trav­el­ers and traders. And we do the same over the networks. 

Malamud: It’s just a dif­fer­ent kind of traveling.

Looking down the road a bit, I’m curi­ous what your thoughts are on Europe’s back­bone, as to whether Europe is going to devel­op a high-speed back­bone, as to whether it’s even real­is­tic to ask a ques­tion like that. What are some of your ideas over the next cou­ple years as to what’s going to hap­pen here in Europe?

Blokzijl: I think in the next cou­ple of years we will see the emer­gence of backbones, plur­al. I don’t believe in this one sin­gle European back­bone that will solve all our problems. 

I think it’s wrong for two basic rea­sons. The first one is from the tech­no­log­i­cal point of view. If you have only one well-established back­bone it tends to be a very con­ser­v­a­tive one. It’s run­ning so don’t change it.

And sec­ond­ly, it’s a com­mer­cial point to view. I don’t believe in sin­gle providers of ser­vices. They tend to be a lit­tle bit more expen­sive than they could be.

Malamud: So you see sev­er­al ser­vice providers, each pro­vid­ing a backbone-like ser­vice. Companies like Sprint or Infonet, or EUnet or…

Blokzijl: Or…whoever. They will come. They will flour­ish. They will go. And new ones will pop up. I think there won’t be that many. It’s I think very inter­est­ing the com­ing few years to observe how many of the tra­di­tion­al European nation­al PTTs will sur­vive in this open mar­ket in Europe. You see prac­ti­cal­ly all the European PTTs are posi­tion­ing them­selves now as inter­na­tion­al car­ri­ers. I don’t think there is room for twenty-five such type of car­ri­ers in Europe.

Malamud: There’s been quite a few people—Geoff Huston from Australia comes to mind, Peter Lothberg—who have been talk­ing about a con­cept of a Global Internet Exchange—a GIX. Such a GIX is a place where any­body can plug in and join the net­work. Do you see that as a nec­es­sary part of the archi­tec­ture in Europe?

Blokzijl: Yes, absolute­ly. If we agree that net­work­ing in Europe is not being ser­viced by this one sin­gle mono­lith­ic back­bone, then we must ensure that if there are three, five, ten, twelve, twen­ty ser­vice providers, that there is full inter­con­nec­tiv­i­ty between them. It’s like the tele­phone sys­tem. You have a plug on the wall, and you are con­nect­ed to the whole world. And you don’t want to be depen­dent upon four providers depend­ing whether you want to phone your uncle in Australia or your niece in Canada. You have one tele­phone at home.

Malamud: [crosstalk]So how do—

Blokzijl: The same with the— So, I think for the user com­mu­ni­ty that it would be extreme­ly ben­e­fi­cial if they define a cou­ple of con­nec­tiv­i­ty points and leave it to the mar­ket to inter­con­nect them. And that can mean that you have sev­er­al ways of con­nect­ing between, or to, these con­nec­tiv­i­ty points. But the idea of hav­ing a few well-defined con­nec­tiv­i­ty points, the jar­gon today is GIX, scat­tered around the world, inter­con­nect­ed among them­selves and from the con­nec­tiv­i­ty points, the local region­al net­works con­nect them­selves, I think that’s an extreme­ly effi­cient and cost-effective way of doing glob­al networking.

Malamud: So is a col­lec­tiv­i­ty point a build­ing, or is it a city, or…what is a con­nec­tiv­i­ty point?

Blokzijl: I think it’s one square meter of floor space with a rack and with a lot of col­ored box­es. Every man­u­fac­tur­er or ser­vice provider will have his box there, and it’s connected.

Malamud: And you just look for some insti­tute like a NIKHEF that’s will­ing to house that con­nec­tiv­i­ty point?

Blokzijl: I think the more log­i­cal places would be on car­ri­ers’ premis­es, because from a tech­ni­cal point of view that’s the log­i­cal choice.

Malamud: It sure makes the leased lines to the car­ri­er premis­es cheap­er if you’re already there, does­n’t it.

Blokzijl: Yeah.

Malamud: Do you see any moves by the car­ri­ers to pro­vide this kind of a service?

Blokzijl: Well, some do, some don’t. There are some US car­ri­ers who have under­stood this and are play­ing with the first instance of a GIX. And European car­ri­ers, some of them—

Malamud: The Washington proto-GIX [crosstalk] is what you’re talk­ing about.

Blokzijl: Yes, yes. yes. And some European car­ri­ers have still to read their first book on the how and why of TCP/IP, so there is a wide gap in under­stand­ing this kind of net­work­ing, and the lev­el of knowl­edge between the car­ri­ers in the world.

Malamud: Well hav­ing them read books of course is a move that I ful­ly sup­port. [Blokzijl laughs] Actually I sup­port hav­ing them buy the books. Whether they read is irrelevant.

Okay, well for appear­ing on Geek of the Week. We’ve been talk­ing to Rob Blokzijl here at NIKHEF. He’s the chair­man of RIPE. Thanks for your time. 

Blokzijl: Thank you.

Malamud: This has been Geek of the Week, brought to you by Sun Microsystems, and by O’Reilly & 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.