Carl Malamud: Internet Talk Radio, flame of the Internet. This is Geek of the Week and we’re talking to Bernhard Stockman who is technical director of EBONE, the European backbone. Welcome to Geek of the Week
Bernhard Stockman: Okay, thank you. Thank you.
Malamud: Why don’t you tell us what EBONE is?
Stockman: Okay. EBONE is a consortium of European network service providers that has seen the need to go together to deploy a common infrastructure for European Internet services.
Malamud: Are you the equivalent of the NSFNET backbone but for Europe? Are you the transit network for Europe?
Stockman: We are not the only one. We are one of them. There are more. Well.
Malamud: Is there a need for multiple transit networks? Do you hit one market and they hit another market?
Stockman: No, no. There’s a need for more. There’s a need for competition. But, EBONE was formed— At the time when it was formed, there was no commercial service available in Europe. So the R&D community saw the need for them to start off to do it themselves, until there are commercial competitors available. And that is now slowly happening in Europe.
Malamud: You say EBONE is a consortium. What does that mean? Do people just throw in resources? Is there a corporation that runs it?
Stockman: No. It’s an association based on a signed memorandum of understanding on what EBONE shall mean to the members. The whole thing was formed in September… Well, the first proposals were drafted in September ’91. And then it was announced for a meeting that everyone that had a network could participate. And that meeting, it was decided that the initial formation of the back bone should be based on voluntary contributions. So that means that whatever resource you had on international networking in Europe, and that you are prepared to dedicate to this effort was so to say gathered into the pot. And after going around the table and gathering all these first participants had, I was [indistinct] down to sit down and think what I could do with all these pieces. So that’s what I did. And up came the first version of the EBONE.
So this was a very informal way of creating an international backbone, you could say. And now when the thing has grown very much and it’s regarded as a rather big success, we are starting to look at a more formal way of cost-sharing. We can’t just go around again and ask for another year of contributions because that means that someone will pay more and someone less. We are now trying to look at fair shares and things like that. So there is a EBONE ’93 budget now based on a fair contribution scheme. So we have gathered everything we had, [indistinct] the cost, and shared among the participants. And the cost is based on the connected bandwidth.
Malamud: The EBONE also provide links to the United States.
Malamud: Has forming a consortium actually reduced the number of links to the US?
Malamud: Is this a regional solution to networking?
Stockman: It’s a hard question, because…you could also say what would have happened if you had not coordinated the links? How many still 64k links would you then have seen across the Atlantic? So I believe— Well, as a matter of fact, in one or two cases, US links would actually be canceled due to the existence of the EBONE links. But the major benefit is of course that a lot of network organizations to date do not have to order their own links but can order links to closest EBONE border system, and by that have the connection to the United States.
Malamud: What type of work is EBONE doing internally as far as coordinating routing and trying to make some sense of the network within Europe? Obviously you have many different countries and many different networks. Is it a tougher routing problem than we face in the United States?
Stockman: Uh…we are trying to make it simpler. I mean, one of the statements in the memorandum of understanding for EBONE states that EBONE shall not hot have any restrictions on traffic. So we are not facing internally in EBONE an AUP problem. We can take on any kinds of traffic.
Now, the routing point you could say is in some sense the opposite of the NSFNET. In NSFNET you have a policy, and you have to make the regionals adhere to that policy. In the EBONE backbone, we have taken out any kind of policy in the internal backbone. And the policy that regionals to EBONE may have has to be implemented in their connection to the EBONE. So if regionals have a policy it’s up to them to implement it. EBONE doesn’t implement it.
Malamud: Aren’t you afraid that you’ll be shattering a bunch of data into some poor regional that doesn’t want it? It’s up to them to provide that firewall and—
Stockman: Yeah. It is.
Malamud: Is that, you think, a general solution to the questions of policy routing and security and other issues in the network, is leave it up to the endpoints? And all you are is a bit pump as a transit network?
Stockman: Well, I’m not so sure if that is the case. I mean, if you are looking at the security I think there was some statement from some source saying that security is not an issue for the transit networks but for the end host. I mean, the security has to be applied at the host. If there’s not sufficiently security management in a host, then it’s very valuable to threats. And I think the same thing goes here, and if you are going to have all kinds of traffic being able to be forwarded by EBONE, then we cannot take on some kind of routing policy restrictions. We have to be able to support all the members, and those are from the commercial, from the carriers, from the governments, and things like that. We have to be able to support all their traffic.
Malamud: Are you in fact a— Do you support commercial traffic on the EBONE?
Stockman: Sure. Sure. Certainly. I mean, the intention here is of course to take advantage of commercial traffic being able to use EBONE. This creates income for EBONE. And by that also is to see as a benefit for the R&D community. I mean if you have a collaboration among— If you see collaboration among R&D networks as useful and good, why shouldn’t that also be valid if you include commercial networks, if you cooperate on the broader level? So that is the intention.
Malamud: So EBONE is a central point where networks can connect. It’s a transit network. It’s AUP-free. This sounds like a global Internet exchange, or at least a European Internet exchange.
Stockman: It has been use— Okay. EBONE has two roles today. It has the role of provision of as I said, [indistinct] in the global interconnectivity. It has also the role of providing backbone capacity to the European countries. Especially for example to the eastern and central European countries where such connectivity is not being able to purchase today. So, they roles may have to be separated in the future. But EBONE being some kind of distributed GIX is a true statement, yes it is.
Malamud: Now, should other regions such as Africa or the Middle East come up with similar solutions? Or should EBONE expand to service those areas? Do you go to Tunisia, for example, or Algeria?
Malamud: And would you?
Stockman: There is a… If you see to day, there is a consortium of European network service providers who have created a backbone. And between there is an agreement that they today shall pay a fee that is based on the connected bandwidth, that is however not valid if you go outside of Europe. Because then you have to see to—for example if you look at US-to-EBONE connections. You have to see that the cost is rather significant, and that there are interests for example from NSF, from [AltNet?] or other US networking service providers to connect to EBONE. So, the agreements—I don’t know if this is formal or informal but anyway, how it is done is that each party here pays its half-circuit. So for example, NSFNET pays the US half-circuit to EBONE, as a consortium, pays its half-circuit to the US. So, the intercontinental links are seen, at least with respect to the US connections, as a common resource, a common EBONE resource.
Malamud: Should there be a regional solution then for let’s say the Mideast or for Africa? Should those countries be learning a lesson from EBONE and trying to form a similar type of consortium?
Stockman: Well it’s possible. I guess if they…have that need. I’m not so sure that they have that need today because the consortium in Europe was formed because of the multitude of uncoordinated links, to try to coordinate them—both internal in Europe and towards other continents. And I’m not sure if that is the case for example in the Middle East and in Africa. But it might be a useful way to start to do networking in these countries, to do it together of course and share financially-heavy resources.
Malamud: What I’m interested in is how do we scale this network up to be a truly global network. And do we do that with a single GIX based in Washington DC, or do we begin forming regional solutions around the world?
Stockman: Okay. That has to be looked a little bit more. I mean, we are trying to deploy what we call a proto-GIX in Washington DC today to test the whole concept out. And if that is successful…okay. And also if there is some kind of— I mean, if you are deploying multiple GIX in the world, who is going to pay the interconnection bandwidth? And what are the rules for these things and— [sighs] If these things are covered, if these kind of administrative issues are really covered, and if there’s an agreement that this is a viable way to do it, sure, it could be done.
We are looking at this as you have to have these kind of neutral interconnects in the world at some spots where all kinds of network service providers have the possibility to interconnect. And this is of course with the intention to keep the global Internet connected. So that we don’t see what some people see as a threat to the Internet, and that is the fragmentation, the balkanization of the Internet.
Malamud: Some people say that a global ATM cloud will solve all these problems, and it’ll provide a virtual worldwide GIX and it’ll in fact do away with the need for routers altogether because you are only one hop away from any neighbor. Do you see that as a trend?
Stockman: No, I don’t think so. ATM could be used to deliver regional services, where there’s a need to multiplex on a lower layer where there’s sufficient capacity to do that. But if you’re going to the intercontinental lines as we see them today, they are fully loaded, mostly. And to put another layer of multiplexing on top of that will not increase the utilization efficiency. So I don’t think that is— I don’t believe it’s a good thing today, anyway.
Malamud: So you think pushing the problems down to the data link is not necessarily a solution.
Stockman: Excuse me?
Malamud: Pushing problems down to the data link, saying, “Well, we don’t need to deal with these problems at EBONE because ATM is going to take care of ’em for us.”
Stockman: I don’t think that’s the way, no. No.
Malamud: EBONE has been very active in using BGP as an internal routing protocol, and deploying CIDR and other solutions. Can you can you give us some insight as to whether these solution such as CIDR appear to be dealing with the problems of routing table explosion and address depletion? Are these feasible solutions to those problems?
Stockman: Yeah. At least— I mean, this has been discussed at length in IETF and I believe for the coming two or three years CIDR will have a…will be a reasonable method of dealing with these problems. And we will of course in EBONE deploy CIDR and BGP before as well, and we will be part of what is known now as the CIDR core, together with some of the US federal backbones and other backbones in the United States.
Malamud: Explain what that CIDR core is, and what the purpose is.
Stockman: Okay, I’ve not been part of this meeting myself. I just very briefly read the minutes. I don’t know very much about it. But obviously, as I think it is defined here, the CIDR core will be the initial top-level transit backbones that will go to decide BGP4 at a very early stage so that later on, regionals behind them can also [indistinct] default route or turn to CIDR, whatever they like. But there has to be a top-level BGP4 implementation in advance of anything else below.
Malamud: So assuming many addresses in Europe fall within a certain prefix that allows EBONE to advertise simply that prefix to the rest of the world and say, “Anything that starts with ‘1925’ is ours.”
Stockman: Yeah. Well in principle, yes. In principle yes.
Malamud: Now is that gonna work or are we gonna end up with so many exceptions and so much fragmentation that it doesn’t solve the problem?
Stockman: Uh, okay. There is always an entropy into this, and again— I mean, again, this assumption that an Internet network will be that stable, at least. That the entropy will not destroy the benefits of CIDR within two or three years’ framework. So, that is what the whole concept is based on.
Malamud: Is it based on an assumption that our growth rate will not increase substantially? Is that one of the fundamental assumptions?
Stockman: No, not just the growth rate but also— I mean CIDR is based on aggregation. And aggregation in turn is based on that all the networks within a certain aggregation area remain within that aggregation area. But of course, networks can move geographically or change service providers or whatever. And that of course destroys the benefits and the possibility of doing aggregation via that network. But the assumption is then that such tendencies will not destroy the overall benefits within two or three years’ timeframe. That is, as far as I know.
Malamud: There’ve been a lot of efforts around the world to coordinate different network activities. There’s been the CCIRN for the intercontinental…what’s it—Committee on Intercontinental Research Networking. There’s been the International Engineering Planning Group—the IEPG. There’s an operations directorate within the IETF. Are any of these bodies successfully able to coordinate the activities of networks, and do we even need such coordination function.
Stockman: Hm. Interesting question. Uh, yeah. [sighs] Yes, there is a need for coordination of networking activities. Because some of the concepts in Internet cannot be…how do you put it…managed from a single service provider, but has to be managed over the boundaries of all the service providers. For example, DNS; for example, address registration; for example, routing policy registration, and things like that that cross the borders of service providers.
Malamud: And these are not just standards-making activities. These are actually operational coordination functions.
Stockman: The standards here are of course for example the BGP4 standard for routing, the DNS standard for DNS servers and things like that that are actually being deployed; that’s the standardization. Then you have operational requirements on standards. That is foreseeing some kind of operational needs in some kind of standard that has to be looked at. Then at the deployment level, you have the coordination of actually deployed software, actually deployed services. And today I don’t think there is… There are some kind of efforts to create such coordination bodies, but no one seems to be really globally accepted today. So we still have a variety of efforts which resemble such coordination. For example in the IEPG; for example in the operations requirements in the IETF; for example in RIPE organization in Europe. You always see— I mean, if you come to the IETF you could as well have an operation coordinations meeting where it’s the same kind of people. The usual victims always turn up and…
I mean…it’s just a matter of frameworks—in which framework you are meeting. So one possible thing would be for example now to invite already-existing service providers in the networking arena, and coming network service providers like the carriers, to have a discussion on what kind of coordination body there is a need for, if it exists, and if not who is going to create it, how is it going to be created and things like that. Even for example compared to the switchboards, you have such coordination bodies. At least in Europe—I believe it’s something called…ETNO…European Telecommunication Network Operators Association—something like that—where there is such coordination among carriers.
So I mean, the question here is where is such an initiative—where will it come from? Should it come from the Internet technical people like me? Should it come from the service providers, from the carriers? Well, I don’t think it’s a big issue where it comes from, but it has to be—it has to come from somewhere anyway. So why not from us?
Malamud: Well the IETF has always been viewed as a very informal American-centric body, and I’ve heard many people argue that in order for the Internet and the IETF and others to be taken seriously there needs to be some form of international accreditation. Do you think the IETF should be located in Geneva with a headquarters building? And should that be a—you know, Internet Society relocate next to the ITU?
Stockman: Wouldn’t that be unfair to the US people to have it in Switzerland? And wouldn’t it be offered to the Swiss people to have it in the United States? Maybe you should look on the moon or something. I don’t know.
No. Seriously. Today, the Internet…the network, I think around 70%, something like that, is within the United States. So I mean… I don’t have a problem with this, personally. It might be that others have it… I think it’s a rather good idea now till we see that of three IETF meetings, probably one will happen out of the United States.
Malamud: Well if the Internet Society’s a global organization, right now the IETF is based on face-to-face meetings three times a year.
Stockman: Uh huh.
Malamud: Is it realistic to think that 700 to 1000 to 2000 people are gonna be able to fly around the world, and it doesn’t matter where you are, a substantial population has to fly in. Are there other ways to do the standards-making process that we do?
Stockman: I’ve no idea. I mean, yeah sure. There is of course a lot of other ways and look at— I don’t know how ISO works but I assume that the complete set of ISO representatives does not meet at once but it has some kind of sub-meetings, separate meetings, and—
Malamud: Oh ISO is very proud to state that on any given day there are nine meetings around the world.
Stockman: [laughing] Okay. I don’t think that’s something to be proud, I don’t know. Depends maybe on the output of— Well.
So…sure. I mean IETF is growing rather big and… I mean, for example we can’t meet at any kind of place today. We have to have special, big hotels and things like that to be able to host such big meetings. And if this tendency grows and gets even bigger of course then course comes again the question of shall we split up, and shall we have for example separate meetings between areas or interest groups of whatever. That’s a possible way of dealing with these problems, yes.
Malamud: Well you’re on the Internet Engineering Steering Group, the governing body. Do you find you’re able to participate effectively in that body, or are you continually getting on a plane or staying on the phone at three in the morning?
Stockman: Ah, am I a doer or am I a goer. Well unfortunately I must now characterize myself as a goer. I’m traveling a lot and trying to organize meetings and things like that. And of course I cannot…put in the amount of work I think the operations area deserve today. That’s true. So my… Well, the area directorship deserve quite a lot of time if it should be managed in a good way, and…I don’t think I have the time. That’s true.
Malamud: Should it be a full-time job? Should you take a leave of absence for a year and enter community service, sort of like the army?
Stockman: [laughs] Interesting idea. Why not. Why not. Community service, serving the standards organization, whoever. Why not.
No— Okay, if not full-time at least something like 25/30% of full-time is at least needed to be able to do the AD work. And for that reason you must have support from your employer. You have to have time able to set aside for this work. Otherwise you can’t do it on weekends and evenings. It’s not possible to do it. So… Well, that the problem with the IESG being a voluntary organization, you know.
Malamud: Is it gonna be able to continue, based on voluntary labor? Is it just getting too big to handle?
Stockman: I mean… It’s not voluntary in a sense today. I mean it— Those people who are asked to go to the IESG I assume go back to their home organization and ask for permission to set aside one, two, three days a week. If they don’t do that, they have a problem. So in that sense it’s not voluntary.
Malamud: It’s voluntary in the sense that you have to serve but then once you’re there you got a job.
Stockman: Yeah. Exactly. Okay, there is no one centrally paying you, so in some sense your home organization is of course funding this. But it’s better if they are aware of it.
Malamud: It certainly helps otherwise they might wonder why you’re never in the office, huh?
Stockman: Exactly, exactly. Or you’re doing a bad job. Because your not giving it the time it deserves.
Malamud: But ultimately do you find that your work with the IESG and the IEPG are beneficial to your employer?
Stockman: Oh sure it is. Certainly it is.
Malamud: What do they get out of it?
Stockman: Uh…I mean…they get out of it a lot of, first of all knowledge of the standards-making process as such and, in Europe that is recently becoming more and more known, right. I mean, a couple years ago IETF was not that known in Europe. It was known in some small sub-communities of the networking people. But now today, because of more and more involvement from European people it’s starting to get very well known in Europe. And of course one [indistinct] serious [indistinct] the Amsterdam meeting that we all— Well, actually shows that there is an IETF Europe as well.
And that is one thing. Now, the second thing is then that European organizations, by seeing the IETF as a standards-making body also have the possibility once they know that it exists and it functions to influence the standards and by that influence their own home environment, in the networking scene.
Malamud: Well thank you very much. This has been Geek of the Week. We’ve been talking to Bernhard Stockman, a technical director of EBONE, the European backbone.
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 firstname.lastname@example.org.
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 email@example.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.