September 29, 2011 - 11:00AM
The following is the output of the real-time captioning taken during the Sixth Meeting of the IGF, in Nairobi, Kenya. Although it is largely accurate, in some cases it may be incomplete or inaccurate due to inaudible passages or transcription errors. It is posted as an aid to understanding the proceedings at the session, but should not be treated as an authoritative record.
>> BERNADETTE LEWIS: Good morning. We are going to start in a few minutes. We are getting our technical issues sorted out here but we will be starting shortly. Thank you.
Good morning, ladies and gentlemen and welcome to this workshop on critical Internet infrastructure which is 555 which is being sponsored by the Packet Clearing House and the Internet Society. And my name is Bernadette Lewis. I am the Secretariat‑General of the Caribbean Telecommunications Union. And we have quite a distinguished panel of guests. I am going to invite Mr. John Curran who is the CEO of the American Registry for Internet Numbers and he is going to give some official welcome remarks to this workshop. Thank you.
>> JOHN CURRAN: Thank you. This is John Curran. When I heard about this panel I had an opportunity to speak with the organizers and it occurred to me a lot of people here think about Internet exchange points but you don't spend a lot of time thinking about the role they play in the Internet and the importance. Internet exchange points come up rarely in sessions like Internet Governance meetings, but as it turns out they are as important as the networks themselves. I have been involved in the Internet since it had more than one backbone. Back in the early '90s I ran one of the first Internet companies. In the United States I ran a couple of nationwide ISPs. I spend most of the time worrying about exchange points than my own network. At the moment we had more than one backbone. We had to interconnect them. In 1993, '94, '95 there was a conscious decision that Internet providers would exchange traffic at locations that were common. We called them network access points or NAPs back then. And everyone showed up at an exchange point. If you think about it, every ISP only serves a very small portion of the Internet.
Every ISP only serves the customers he is connecting. Even the largest ISP in the world might only be 5% of the global Internet. Of the other 95% of the traffic that goes out his network leaves at exchange points. So, in fact, the Internet is gated on exchange points, the capacity and the performance of them. Because every Internet company serves such a small fraction of the total Internet and the rest of their access goes via exchange points. They are where ISPs come together and meet and they meet in a common fabric where they share traffic on one particular network or they go to an exchange point and they exchange traffic directly with their peers. Exchange points grow up where ISPs need to exchange traffic and that's been the classic evolution of it.
You have a lot of traffic between two ISPs in a particular city, whether that's Amsterdam or whether or not that's New York, an exchange point appears. When you talk about the Caribbean and developing nations, this topic turns out to be more important. When you can't exchange traffic locally the place you are going to exchange it could be on the other end of a very expensive link or the end other of another link. So the fact of the matter is that the ability to have a high performance Internet in any part of the globe in any region is predicated on having exchange points. Most exchange points have evolved over the largest providers. Say we have too much traffic in a city. We have too much traffic in London and we need to build another exchange point so we don't have to carry it all to Amsterdam. They may not get to developing regions of the globe and say we don't have too much traffic. The reality is they don't have a lot of traffic. That doesn't recognize the importance of an exchange point locally in terms of performance and in terms of having reliable connectivity. So this is a huge issue.
The fact of the matter you cannot let it be driven entirely by the carriers where they want exchange points. If you want reliable service you need to think about the fact that this is a key part of Internet infrastructure. It is not that difficult to operate. And it is one that developing nations can do on their own and actually attract improved infrastructure to their area. I am pleased that there is a panel at the Internet Governance Forum to talk about this. I think it is a great panel. I hope I have helped explain why this is important to you folks and I wish you the best on the discussion. Thank you.
>> BERNADETTE LEWIS: Great. Thank you. At this time I would also like to just welcome our online participants. We are really pleased to have the opportunity to engage remote participants in that very important topic. And John would have given a bit of an oversight in to the way we regard Internet exchange points as critical Internet infrastructure for emerging economies and those seeking to migrate to knowledge societies and Internet based economic activity. So this is ‑‑ in this session we propose to explore the key issues surrounding Internet exchange point proliferation in developing countries and we have quite an impressive panel of practitioners who have been engaged in different aspects of the Internet, the Internet economy, the establishment of Internet exchange points, traffic routing, regulatory and policy implications for emerging countries which are seeking to really capitalize on the benefits that can accrue from having an indigenous local Internet exchange point. And our speakers today are Taylor Reynolds of the OECD. He is a global traffic routing expert. Then we have Bevil Wooding from Packet Clearing House. He is an international strategist very actively engaged in IXP promotion and proliferation in the Caribbean. Packet Clearing House is a partner with the CTU in the establishment of Internet exchange points in the Caribbean. And finally Michuki Mwangi, he is a senior development manager for Africa in the Middle East of the Internet Society and he is also very intimately associated with the establishment of Internet exchange points in the developing ‑‑ in developing countries.
This morning we propose to discuss the international traffic routing practices, domestic routing practices for Internet traffic, the role of an Internet exchange points, how you engage Stakeholders and how you mobilize them to get the buy‑in to foster cooperation amongst competitors, pretty much competitors in the same market space, to work on the establishment of an Internet exchange point for mutual benefit. But I think just given our experience within the Caribbean we put forward the notion of having Internet exchange points. So Internet exchange point proliferation in the Caribbean is an issue of national development. If that's the underlying principle we are seeking to foster national development it takes on a whole different perspective and we will be exploring some of these things this morning in this panel. I don't want to take up too much of our panelist's time. I am going to invite Taylor Reynolds of the OECD to start the discussion this morning. And he is going to be talking about the economics of an Internet exchange point.
>> TAYLOR REYNOLDS: Thank you. Okay. Thank you very much for the chance to participate with you. I want to start out by saying that I am an economist and I am trained as an economist. So what I talk about today will be from an economist perspective and I think we need to have economists on board to understand how these things work together. If you really want to get an economist excited there is one word that sums it all up and if you could advance ‑‑ oh, it is here. The one word, let's see, markets. So economists love markets because everything is dictated by markets in economics, almost everything and markets are placed where people come together and they exchange things. They exchange goods and services and they either do it for money or they barter. They are central to the functioning of society. Internet exchange points are fascinating to economists because they are markets for the transfer of data. So when we talk about Internet exchange points we do it in terms of markets. I have been in Kenya now for four or five days and we did a trip out to the Maasai Mara. I saw a lot of people walking distances to work or to sell their goods. What you can see is that distance to the market matters in terms of their life and in terms of how effective the society is and if they have to walk long distance they spend way too much time getting there and not enough time exchanging. So keeping in mind distance to market matters that is essentially what we are thinking about in terms of domestic Internet exchange points. If we have Internet exchange points that are too far away the distances are too long for the market to be very efficient.
It is a chart that shows the number of countries that do have an IXP and the number of countries that do not have an IXP and you can see the dark blue is the area that has not lit up so to speak and a majority of countries do not according to latest data. There is a lot of work to be done in order to have traffic exchanged locally. Now I want to talk about what are some of the issues in a country that does not have an IXP.
Well, first of all, Internet access is expensive. So and again there is a lot of technical people in the room. This is more of an example for the nontechnical people. When you do not have an Internet exchange you end up having to send your data traffic outside of the country. You transfer it to another provider and it comes back in. So they call this tromboning because it looks like the shape of the horn, the trombone and what I have shown up here is someone who is exchanging traffic and they are from Mali and they might be on one IXP and they send an e‑mail to a friend who is a subscriber on another IXP. IXP A, and IXP is not locally and it has to travel up to France and it is exchanged up there and travels back down. You have to pay expensive international transit. You are paying it twice. Those costs have to be eaten somewhere. And they are passed on to the consumer. So if you do not have a place to exchange traffic locally, your Internet access is going to be more expensive.
The second problem you have is latency and this is for those who are nontechnical. Latency is that lag you get when you are trying to talk to someone far away or when you try to download something on the Internet and it takes longer. You will get slower response times because of distance and because you are traveling over several different hops to get up out of your country, from Mali to France and then back down, there is increased potential for international congestion. So you really are having a slower reaction on the Internet because you are not exchanging traffic locally.
So one of the things that I found fascinating about IXPs is they really focus on cooperation as a model. This is an area where competitors should get together and find a way to reduce costs for everyone because that benefits everyone in the society. So local operators can significantly reduce their transit costs internationally if they come to an IXP locally and exchange traffic. And also one of the nice things about having an IXP locally is that it reduces the barriers to entry in the market for other firms. It can help sprout the evolution of a data center and it can help other companies jump on to the Internet without having to buy very expensive transit. It also will improve the experience for users. And like I say if you are trying to access content locally and it is taking a long time to download, you will have a much faster interaction on the Internet if you have locally exchanged traffic.
Okay. Finally, I will leave the rest of this to the expert, but I want to talk about what is needed to have successful Internet exchange. Now a lot of people who come in to this very early on say well, the key to having an Internet exchange is you have to have servers and an Internet connection. So I think this ‑‑ the first reaction is the solution to this is equipment. And actually I would put equipment very low on the list in terms of what we need because you can always get equipment. What you need to have, first of all, is a vision. Someone has to have an entrepreneurial spirit come in to the country and say this is something that you need or that we need here locally. And put this in to practice. They also have to have a long term view. This is not an I show up in Mali. I install some equipment and then I leave. You have to have someone there who has a long term vision of growing the Internet. You have to have people who have technical skills and also skills for managerial tasks and sales and marketing. Someone who starts an Internet exchange point they need to be an excellent marketer because they have need to convince competitors to come in and exchange traffic locally. You need infrastructure and this would be equipment and servers and also infrastructure such as electricity. You need to have stable electricity supply in order for the exchange to run. One thing you need is competition in the local market and the reason for it if people cannot get to the exchange via a leased line you are going to have a difficult time having an exchange. It is extremely important. (Off microphone). So with that I say thank you.
>> BERNADETTE LEWIS: Thank you, Taylor, for that excellent introduction. I certainly love the analogy of the marketplace and we see the benefits already. They reduced transit costs, reduced latency and improved experience for users, enhanced efficiency of network resources and, of course, the opportunities for business and more common in the information and the traffic that is being passed.
So now Taylor has very establish ‑‑ established a very good overview and an introduction to the topic of Internet exchange points. I am going to ask Michuki Mwangi to talk about his practical experiences in the establishment of Internet exchange points in emerging societies.
>> MICHUKI MWANGI: Thank you.
>> BERNADETTE LEWIS: They are going to just be exchanging here. While he is setting up I wanted to speak about the skills to which Taylor referred and our experiences in the Caribbean. The technical part of the establishment of an Internet exchange point is the easiest part. Is that fair? And the ‑‑ you really need to manage relationships. There is a lot of social engineering. How do you bring competitors and in the Caribbean it is not just competitors. It is antagonistic competitor. How do you bring them together to work on an Internet exchange point which is of mutual benefit. We pitch it as an issue of national development. Okay. Our next speaker is ready.
>> MICHUKI MWANGI: Thank you. So I think I'd like to pick up from where Madame Chair left it. The challenge of setting up Internet exchange points basically starts from the fact from the experience we have had in most developing countries is that the technical takes the least time and least effort and the social side takes the most. So we have a saying that setting up an exchange point especially here in Africa is 80% social engineering and 20% technical engineering. And because of that the question is how do you go about getting the exchange point set up. One of the things that we have tried to work on is getting the people to sit and talk. The competitive nature of Internet access here in Africa, one of the challenges from the regulatory environment and then secondly from the competitive nature in that it is a small market. It is a small nation and we are trying to get this number of people on the net and get your time and investment as a business. Running that business as an ISP, it is a small market and it has been for a long time and for that reason it created environment where there is a lot of mistrust within the community and within the ISPs.
So I remember the time when I used to work in Kenya and used to work for an ISP and it was almost criminal for me to be seen talking to another engineer from another ISP and I would actually risk losing my job if that went back to the management. That's how tense it was. And today I have drinks with them and it is fine. As it is if whoa. If you have that kind of environment where even engineers are not allowed to talk to one another and how do you then get the senior management to sit down and agree to connect to an IXP. As a result of that you will see that the different environments across Africa will actually result in different models coming up, being set up on exchange points.
Now what's happening in Africa? Well, there are 24 Internet exchange points in Africa. I think I should say 25 now. And because we have one recent one and I will talk about that in a moment. Of these 24 Internet exchange points we have tried to classify the levels in which they are at. We found in Africa we have places where they do not have Internet exchange points and some places they won't have Internet exchange points in the near future because of the policy regime exists. You need at least three providers in a country for there to be an Internet exchange point. And then we found what we call the level 1 exchange points which is basically boxes and wires and lights showing up. And we have seen a lot of interesting cases where nobody even knows where the padlock is to lock the exchange point and nobody knows what has happened, the exchange point has gone off and basically it is a volunteer basis of who is closest to the exchange point and who has the time to make sure that everything is working.
And others who actually plan to have an exchange point. And when you go it is set up there is a switch, probably one or two routers but no packets have been exchanged. Just boxes and wires and then we have level 2 which is basically core functionality and this is where you find there is an exchange point which is exchanging traffic. At least if you send an e‑mail you get a reply from somebody and they will tell you well, I actually don't know like what is the statistics of traffic being exchanged across, but I do know we have these number of members and some of them can actually ‑‑ are sending traffic. I can tell you what my company exchanging is but the traffic we are exchanging on the exchange point but I don't have a (inaudible). Sort of trying to get their act together and provide basic core functionality which is exchanging data but beyond that you struggle to get additional information or services at that particular exchange point.
And then we have an exchange point which is at level 3 which we say is catalyzing growth. This is an exchange point that has gone beyond a Website of having information and is providing additional services. You can go to that exchange point and find DNS root servers and go to that exchange point and find a network time server. You can go to that exchange point and get value and it becomes attractive and the members participating in that particular exchange point are not just ISPs. You will find diversity in the participation. You will find academic networks. You will find governments connecting and then you will see that at that point it starts becoming more critical to the community and that's where we are looking at the level for IXPs. And these are the ones that have managed to get to a point that everyone is significantly dependent on the operational stability of that facility.
Governments get interested in that exchange because if the exchange point goes down we get affected. But even more interesting, it is here in Kenya, the Government came to us and said this facility is really critical to the day to day operations of some of our agencies. It is paramount that you provide measures to put in place to ensure that it continues to operate 24/7. So at that point we can say an exchange point is driving and has reached what we can refer to as a critical infrastructure.
So and to start with you will find that when an exchange point starts and as my colleague who spoke earlier, Taylor Reynolds mentioned that when putting up an exchange point is really ‑‑ the difficult part is getting people to talk. Once you get people talking, and agree the technical bit is easy to get done. That takes 20%. You have a long period of time where you spend getting people to sit down and talking. And we had a most recent experience where I will say this is the fastest one I have seen is in Lesotho. It is a small country inside South Africa and the challenge they had they were paying an extremely high price to get international access through South Africa. It is a country within a country. And then the costs of sending traffic via South Africa for local traffic was getting to them. So it took them roughly one and a half years to agree. Took them about one and a half years to actually agree and, you know, the community sitting down having meetings every now and then to agree on the structure and the model of the exchange point and where the exchange point is going to be located. Couple of issues on neutrality and it is now located at the National University of Lesotho. And it only took us about two months of planning, getting everything together, logistics in place and their minimal investment and one week of hands‑ on training.
One thing that was mentioned you need to make sure that when the exchange point is up and running you will not experience what we have seen in some countries, the boxes and wires have been put up and if there is a network routing problem people go to the switch and remove the country until the network is fixed. And once it is fixed you put the cable back on the switch. You want to make sure that the engineers in that community are able to run and maintain the exchange point beyond the setup. So we facilitated, the Internet Society, a training, five days hands‑on training and we launched the exchange. Up there I have a graph which was taken just two days after we set up the exchange point about a week after. That's just two of the top providers connecting and exchanging traffic. It is not an exchange point. This is the incumbent and the next competitor exchanging. They had peaks of almost half a meg of traffic being exchanged. The population is less than 3 million people in this country. That's the starting point.
So the phase that goes past this now. We have the boxes and wires working. Now how do we get them to grow to a point that they actually are a thriving Internet exchange point. Now that's where we start looking at bringing value to the exchange point. By bringing that value it means that the participant of those participating in the exchange point are going to drive more value. At the exchange point they can get something out of that particular facility. And so I am going to look at another example of the exchange point I mentioned earlier and this is the Kenya exchange point. I put up the statistics which is more or less the period of today, just a few minutes ago and over the past one week and you will see that the traffic is, you know, compared to what we have just seen about the Lesotho, this is one gigabyte of traffic being exchanged as of now. So that's just here in Kenya. And the question is how did we manage to get there. And the answer to this is that the community here has to be consciously involved in understanding what it ‑‑ what the exchange point is and how to bring more value to that particular exchange point. So to start with, we have Government appearing at exchange point and there is e‑Government content available. We have content providers, that means non‑ISPs but we have people who can actually bring content. We have the academic network which is providing content to the universities. And now as more people learn to appreciate the value of an exchange point we have managed to get some providers offering to host and serve Google cache and ACME cache through the Internet exchange point. It has become attractive to a lot of people. Anybody who is in this region is interested to connect to the exchange point. You can get access to Google cache. We actually are in the phase of trying to test ACME, ACME caching as well at the exchange point. So we expect to see this traffic growing.
The last time I looked at statistics maintained it showed that Kenya was growing at astronomical levels just because of introducing Google cache and other content providers. You need somebody who is seeing the vision, a long time vision of what the exchange point needs to be for it to grow. I would like to add on to that and say it does not need to be an individual but it needs to be the exchange point as an entity, because when you depend on an individual they can grow tired and the job can get the better of them and then you see the exchange point not growing. You want to see a well‑ established framework. And trying to get the members aware and how can we bring value to an exchange point and how can we bring value so that we also benefit from the exchange points and each member will have to contribute to the growth of the exchange point.
Now just to conclude, one of the things that we have started observing and this is based on some work we have been doing in Mozambique, it is very easy to trigger their thinking of how to get value out of an exchange point. Now in the time we spent in Mozambique we are trying to do a study to understand why is the exchange point not growing as well, as fast as what we call private. That means that people are going off the exchange point to peer private and our understanding is that in addition to the fact that they have challenges with infrastructure, the value that you perceived to gain from the exchange point was diminishing.
So how do you make them realize that the value that is diminishing is by the fact that they are not contributing to the growth of the exchange point. It was an interesting discussion that when we asked them, some of the ISPs in Mozambique are connected to several neighboring countries. So how do you get them to say you can actually bring value; you are connected. You have an office in South Africa. Why aren't you bringing the traffic from South Africa in to your office. Why aren't you reselling the traffic from the exchange point to bring regional activity within the exchange point. You have the value to bring to the exchange point. Another one was selling ‑‑ one of the key products that they provide is anti‑virus software and stuff like that for their customers. I say well, you do not need to sell it to only your customers. Sell it to the customers of other ISPs. Just make sure when they come to update the virus database from your servers it goes through exchange points. They will still buy your products. So with a little bit of discussion we realize that at the end of the project we had so many people rethinking how they can actually gain value out of the exchange point. Now how do you calculate that value? If you look back at this graph, this is one gigabyte per second. If the exchange point disappeared all that traffic will need to go somewhere.
And wherever it would go it means it will not go through the exchange point where everyone is interconnected with one another at no cost. It will have to go up through their upstream providers. Now if you calculate at the rate of about $300 per megabyte times one gigabyte, that's the cost of each ISP will have to be shared across all the participants in the Kenyan Internet exchange point and that's somebody who has to pay. And if you look down at how it trickles down that means that the end user has to pay for it. That's a cost savings that goes to the members and trickles to the end users who access the services.
Now here is something else that goes to the next level and I had sort of mentioned it in my previous discussion. So when you have an exchange point that is operational what is the next phase when you get that critical level? Then you start looking at the regional aspect. So we are interconnecting traffic within the country very efficiently. So what next? The next phase is let's interconnect across the region because in Africa and this was happening until when was this? 28th September on the right‑hand side. 28th September at 1600 hours, just right before 4 p.m. From Tanzania to Nairobi it was 400 milliseconds. You can see between ‑‑ it was between 200 and 400 milliseconds. That's a trace route from Tanzania to Nairobi. Tanzania is right next door and we have multiple ways of getting to Tanzania, represent through the submarine cables. But we actually have to go to London to connect to Tanzania and back. And now this is a trace route from one of the providers, Cybanet who realizes there is value of getting them connected directly to Nairobi. They are a service provider in Tanzania connected to a point. Now what they have started doing is making sure that their routes are visible at both exchange points. So when they trace routes from Tanzania they don't have to go to London but come to Nairobi directly. They are saving their international capacity for regional traffic.
Now what we will start doing in the future is start measuring what percentage of that traffic is going across the region. Once we are able to do that we are going to the next level of not just saving international costs for the local level but also other regional level. These are still challenges that we need to address but that's progress. And a lot of this is being triggered by a Forum which happens every year. And it is an initiative from the Internet Society which is basically aimed at bringing ISPs together to discuss the issue of what is the value, what is the business and economic value of me interconnecting to an exchange point within my region and within my country and within the region. With that I would like to stop there. Thank you.
>> BERNADETTE LEWIS: Thank you very much for that excellent overview. One of the things I wanted to say despite the level of classification of the Internet there is benefit even at level 1. I don't want to stay at level 1. It is a progression that you want to make a process of continuously adding value and, of course, there is the need to educate the Stakeholders and keep them involved in the process. So I think that education is awareness is also key to this entire process. If I could just address the online participants, do we have ‑‑ are there people who are following us on online?
>> Yes, we have a Fiji. We have Palestine and we have Jane Pool.
>> BERNADETTE LEWIS: We thank you for following the discussion. We are going to move on to our last formal presentation. This is Bevil Wooding and he is going to be talking about the experiences in the Caribbean in terms of establishing Internet exchange points. Bevil.
>> BEVIL WOODING: Good morning, everyone. As one of the organizers of this workshop I thank you for your interest in what we consider to be a very significant subject. I will be sharing this morning on the experiences of Packet Clearing House in working in the Caribbean and raising awareness about Internet exchanges. So let's get straight in to it.
One of the things that we have seen over a ‑‑ particularly over the past two to three years is an increasing awareness of the requirement to do something different about how infrastructure and the approach of nations towards ICT based development is taking place.
And so we are hearing comments, and I am paraphrasing here, there is a growing need to leverage technology to educate communities, develop economies, secure citizenry and advance societies. This has largely been up until very recently cliched speeches. There started to be a new interest in what exactly is the approach that we need to take towards this knowledge base economy and building these knowledge base societies.
And so as people realize the movement towards more e‑everything, e‑applications, e‑Governments, e‑learning and so on they started to realize we are facing a fundamental problem in the regions. Even though we are putting applications and services and information online we are not getting the kind of performance or value we would expect when we look at models in the developed world. There was a recognition, something has to be in place to make it more reliable. As countries move towards these applications online and with an Internet face start to realize there is greater exposure to cyber threats and the Caribbean regions because of the vulnerability of the networks is a target for significant cyber attack and that has played an interest in for some mechanism to improve how domestic traffic, data and information is protected online.
Greater focus has to be placed on education and awareness. There is a need for new levels of collaboration, particularly between agencies and between countries but at the same time there has to be a move towards building indigenous capacity and improving the policy and legislative frameworks that govern and determine how these e‑based application services could be ruled out. And for this discussion a recognition that there is urgent need to improve and strengthen physical infrastructure in the Caribbean region.
And so that has led to Packet Clearing House working with organizations, most notably the Caribbean Telecommunications Union, ISOC to provide what this model shows here, initiatives to address these issues, education and awareness, collaboration and capacity building and working with regulators to develop appropriate frameworks and, of course, infrastructure strengthening, dealing with the issue of traffic exchange in the region which has been discussed by both Taylor Reynolds and Michuki. They are realizing as more services go on outbound cost of introducing Internet based applications is also growing. And these are economies that can't afford this kind of economic hemorrhage. And essentially what has been termed as subsidizing development in developed countries from a developing region. So this is the project we have taken and it looks at and recognizes that there are several parallel tracks taking place. We know we are not working in an environment where the whole focus is Internet exchange points. With the growth of the Internet and with the growth of Internet access and increase in broadband services there is a proliferation of indigenous content. There is growing amount of Caribbean content. Most of it outside the Caribbean region, emerging communities of interest and advocacy group. Operators group started just over a year ago and is now making approaches to bringing together the regions, technical practitioners and similar to the example that was given earlier you had scenarios where operators or engineers in one ISP or one telecom operator would not be seen relating to in public operators or engineers in one region.
At the same time through the Caribbean ICT road show which is on initiative design to raise public awareness across all sectors for the need for ICT based development this initiative has been a major platform in our being able to get the audience and the interface and discuss the issues of domestic traffic exchange in the Caribbean region. And this is a signature outreach to Government ministers, to policymakers and regulators and to youth and educators and business leaders and this approach meant when we introduce the notion of there is need for domestic Internet traffic exchange we are not talking to the technical folk. We are not just making it a technical issue. What has led to the success that we have seen over the last two years is that we have opened it towards to being a national development discussion and the social advance conversation. And this has proven to be very, very successful in getting the attention of the operators. Because now essentially to say I don't want an Internet exchange point is to say I am not interested in international development.
And so we have found a way to hold them to their word as it relates to I have a Stakeholder rule in national development and I have a part to play. So help us to advance them in the countries that we are working in.
And then finally the build‑out of Internet infrastructure has not just been let's get Internet exchange points established but put all the services in place to create a robust Caribbean Internet economy. That means let's talk ‑‑ let's get the root servers. Let as port the content providers. All of these things have been taking place against this backdrop. New groups of users, technical communities bringing up increasing public awareness and the build‑out of proven infrastructure. A lot of telecom providers, for example, already have an imperative to improve their infrastructure because of the proliferation of mobile devices. The Caribbean region boast greater than 100% mobile penetration. You have mobile users who are making demands for Internet services and improved Internet services via their mobile providers. It is to improve the efficiency of Internet service delivery across mobile devices. That has provided a wonderful backdrop for us.
I want to close by going through some of the issues that are still at play. This is the map showing the cable, submarine cable systems in the region. The region is well covered, well served, well supplied in terms of submarine connectivity. The irony is most of that connectivity capacity does not translate in to the kind of broadband services that you would expect in a country. Cost has to be brought down and that speaks to some of the competition issues that were raised earlier. We have largely markets that have two dominate players wrestling them out and the prices seem to not reflect what you would expect to see in a competitive environment. And this is a real issue. When you look at this map you would see there is a lot of opportunity for capacity delivery to the countries, but we still need to see that reflected in price and access in country. Transit relationships, most providers still buy international transit. Most traffic is typically exchanged in the Americas, in Miami, in the United States and there is a significant amount of interisland traffic because of this.
And so using the same tromboning map absence of ISPs in the Caribbean compromises our ability. When you look at the information growth of Internet services you see why this situation can't be allowed to continue. Pairing relationships, let me pull up this map that shows the two dominate, the two 500‑pound gorillas in the Caribbean Sea. Between the two major Caribbean networks there is no direct traffic exchange and the entities between the Caribbean regional service providers are all exterior to the region, Sprint, Telenet. That's where the traffic exchange is taking place. This research was done by (inaudible), one of the first research projects coming out.
So you see that the situation even though you have significant traffic being exchanged in their networks in Columbus and cable and wireless and that traffic is not being exchanged between their networks directly in the regions. So where Internet exchange points in the Caribbean, I call this my map of our progress. Right now we have active Internet exchange points in Carrizal, Haiti, most recently in the British Virgin Islands and in Grenada. In progress we have discussions going on in St. Kitts, Dominica and St. Lucia. Some are saying they are thinking about it, Jamaica, St. Vincent in active consideration mode. You can interpret that as generously as you wish and everyone else is saying maybe one day. Let's see what will happen in the other territories before we make any kind of investment in it. And so this is a map of where things are at in the region right now. We have learned several lessons over the last few years. One, needs to engage all Stakeholders not only the Internet service providers. Two, present the case in clear terms. We found it, for example, useful not only to say here is a situation in Africa or Europe or America. We had to get the Information Internet Traffic Exchange and what was it costing the territories and region to make a case for the Government. You are hemorrhaging foreign exchange in an environment where you need to hold on to foreign exchange.
So we to get the fact not just generic justifications this had to do with the reality that we are not dealing with ISPs but open the debate to the nontechnical community. So we had to present the case in terms that they can relate to and understand. Third, the need to follow through. We found very early on that it was not sufficient to simply say this is what IXP is about. You need it. It is wonderful and will do you well. We had to follow through on the environments to see what we can do to help you to move along. Whether that is technical advance, facilitation amongst fierce competitors. And then finally communicating the notion that the IXP is only the beginning. Countries, territories have to build on it. We have to bring value and we have to share the experience. And that's one thing that we are still working with the existing exchanges in the region to capture statistics and get data and how is that translating in to new economic activity in the marketplaces. With that I think thank you and I pass it to the Chair.
>> BERNADETTE LEWIS: Thank you very much. If I could just add a couple of little things to what Bevil mentioned. Education, you cannot underestimate the importance of education and raising awareness, bringing the issues to the table and engaging the Stakeholders. And one of the things we found particularly useful in the Caribbean was the engagement of the ministers responsible for ICT and they provided a tremendous impetus in getting the Internet exchange points up and running, because when we presented the information, for example, at one of our outreaches in Carrizal it was the minister who dictated that we needed to have an exchange point in Carrizal and similarly in Grenada and also the BVI. Having the political, the policy level that's sort of support is a tremendous impetus but really it is about bringing the Stakeholders together. It is about engaging them, allowing them to come up with decisions that are right for them and facilitating a process. And with that I am going to open the floor now to any other ‑‑ any of our audience who would like to talk about their experiences with Internet exchange points. We have someone from Cuba first to ‑‑ three. Great. Thank you.
>> Thank you. I, first of all, want to thank Bernadette and the panelists for the excellent presentations. Almost all has been said, but nevertheless I want to ask the panelists to please if they can to expand on three subjects. First subject is one of cost. It has been said that one of the potential cost savings of Internet exchange point is the possibility. Aggregating traffic of different ISPs. This allows you to negotiate better agreements with tier 1 providers. Sometimes to get passing this agreement from transit agreement to peer agreement. I want to ask the panelists if they have some experience about this happening some place and expand about this.
The second thing that I want the panelists to expand a little bit is that it already has been said the IXPs is the first step but you need to add value by hosting. It was mentioned that in some of this IXPs has been hosting some services like the Google and ACME cache but it was also mentioned the need to put root servers, but hosting content, content that is designed for local consumption from e‑Government or e‑education. And I'd also like to ask the panelists if they have some informations regarding this or best practice if this has been done in some of the IXPs.
And finally the third area which I would like the panelists to expand is, of course, that the next step is regional IXPs. The potential and the value that regional IXPs as an extension of this technology, this has been mentioned already. I would also like to ask if the panelists, they have some experience in some areas of the world in which regional IXPs has been established or been negotiated. And please if they can share with us around this. Thank you.
>> BERNADETTE LEWIS: I am going to ask Michuki to address the issues. I think you covered some of those in your presentation. Expanding on the value and regional benefits that can accrue.
>> MICHUKI MWANGI: I think the first question, the first area of elaboration if ISPs have come together to negotiate for cheaper capacity for upstream providers. Well, there has been efforts especially here in Africa to try and do that, but I think the model that ISPs operate in tend to create more tension on or sort of lack of a middle ground for them to come together than there is the opportunity for them to come together. And the questions that people tend to ask mostly if you negotiate for, you know, as a unified group to buy the capacity, you know, how do you work out the billing model because you know it is not that individually each person of the build but you are looking to buy bulk capacity. How do you work out that expansion. I remember here in Kenya we tried that at the end of the monopoly that ended with the incumbent around 2003, 2004. ISPs tried to form an association to set up their own gateway and buying capacity, but that didn't work really well. Mistrust. Everyone felt they would get a better deal by going to somebody else. That mistrust tended to be quite a big factor in trying to get this result. And as a result if you look at the way exchange points end up being said they are often neutral entities away from the ISPs such that they would prefer a neutral entity and even the location of the exchange point has to be neutral because the perception that ISPs will always have is that if it is located in one of our members facilities that member will have undue advantage over the rest, but in the reality of it all it is ‑‑ really that's not the case but it is, but the environment that creates those perceptions but not the hard reality that comes to the operational aspect. The other point hosting environments, one of the things that is important for exchange points they should never engage in a business that interferes with a potential member.
Now if it starts going in to the business of hosting it means that they are killing a potential business of a potential member and that in itself will start eliminating the exchange points from its membership. And so it will be moving away from being an exchange point in to being a competing business. So that's often not encouraged. However you can host content that brings value and that's why we were talking about DNS servers and maybe a Google cache if you can find a mobile but you are bringing value and not providing business. There has to be a distinct and clear policy with respect to that. The last thing was on regional exchange points. Now there has been a lot of ongoing discussion about what really is a regional exchange point.
Now if you look at the way exchange points are built, what makes exchange points attractive and successful is the membership and the value you get by participating on that exchange point. If you go to an exchange point and there is no traffic, exchange less than one megabyte of traffic there is no value. If it is valuable then it will attract more people. I will give you an example. If you look at the London Internet exchange point do you call that the UK or British Internet exchange point or a regional exchange point or a global Internet exchange point? It started as a national exchange point and became a regional exchange point when other people from other countries across the neighboring countries came to peer them. And then people from other country continents started coming to London. We have operators from London, Latin America and from Asia and U.S. and Australia. It has gone from being a national and to regional.
We are going to go and build a regional Internet exchange point there because at the starting point and if you look at the graphs I showed of Lesotho only the value they bring there is what will be attractive to the people participating for other people to come and join but I ‑‑ on a national exchange point can evolve from being national to regional to global based on the value it brings to itself as it attracts as the years go by. It is difficult to wake up one day and say we are going to build a regional exchange point in this location. It is difficult to wake up one day and say we are going to build this massive airport hub. There must be enough incentive and traffic attraction for that to happen. I think I have answered all the questions.
>> BERNADETTE LEWIS: Thank you. I really love the evolution, you know, of the Internet exchange point. We had a question. Could you identify yourself please?
>> TROY ETULAIN: My name is Troy Etulain. We are helping Afghanistan set up an ISP. And I wanted to point a few things out that the process experiences has helped us understand and the IXP in Afghanistan is going to be set up in the same building, same room as the exchange point between the mobile network offices. So for that fact one thing that hasn't been pointed out yet it is not about mobile Internet. It is about all mobile services. So to extent that mobile network operators can route their traffic via the Internet to towers. So ‑‑ and I was originally driven to a mobile Internet system of IVR. Achieving efficiencies for regular mobile systems. The other thing that I think that would be really nice to hear the panel respond to would be this question of governance of. I think that Colleague Michuki touched on this briefly but I think as IXPs begin to lower costs, increase speed human behavior and businesses follow that. So then it becomes potentially possible, for example, online servers with video to become sustainable and popular and accessible and influential politically speaking.
Once that has been achieved then the IXP services its functioning has a political dimension to it, because like in Egypt they might want to shut it off if it is functioning successfully. A technically economically successful IXP raises questions about a what kind of national asset. What about sovereignty in which sovereignty questions are raised by the fact that we have a more dynamic and affordable Internet. It comes across as a content question. I wonder if you can address that.
>> MICHUKI MWANGI: Because that actually is the flipside of the coin when you say let's make IXP a national issue. Flipside is how it takes the attention in the national Government and national community and it places a different priority on who governs and who watches over and who controls this Internet exchange point. We have largely promoted the open and mutual model. What you describe is a scenario that comes with the Internet exchange point, grows and starts to change things in the markets. And we haven't gotten to that stage in the Caribbean. Even in the formation of the exchanges Governments and regulators taking a role and saying this is a priority and your service providers and telecommunicators move forward with it.
There was two interesting models that were taken in. In the case of the British Virgin Islands the regulator had to issue an edict. They were saying to the service providers I need you to engage in domestic peering and they made it an offense under the Telecom Act to exchange any domestic bound traffic externally. That was not an advisable approach. We didn't want to see the hand of the regulation coming in. And they felt in that environment there was no other choice. The providers were not prepared to come to the table and discuss traffic exchange. And one told me directly this they had better things to do with their time.
In Grenada the threat of regulation was sufficient to get them to the table. If you don't do this and if you don't do it in this time frame, I will be forced to regulate and that was sufficient. Got together and Grenada's exchange point was set up literally months after that command was issued. You have this situation whereas it becomes increasingly important to not just Government service delivery but general service delivery. We anticipate there will be a different kind of scrutiny. Regulation is about development. It is a hard thought for some and that's the conversation we are having right now. There are more services that now depend on this facility. Those who are responsible for governance must take a more mature look at it in making it stable and reliable.
>> BERNADETTE LEWIS: Thank you. Yes, one in the back and then Akeel. And then we take any comments that we have online. Would you identify yourself, please?
>> NURANI NIMPUNO: Thank you. My name is Nurani Nimpuno. I work for Netnod. We are the oldest exchange in Europe and we are a root server. First of all, I would like to congratulate ISOC for the fantastic work they are doing in Africa. They are making an impact in Africa. And if there is anyone you need to talk to get ‑‑ find out more about exchange points do grab Michuki because he is the expert. I wanted to ‑‑ I think in the last question Michuki actually touched upon the neutrality aspect and it is something that is very important. So, for example, when our exchange point in Sweden many years ago was set up it was actually because the two large telco providers wanted to interconnect but they didn't trust each other. And that's actually where the academic network stepped in and said well, we can put a switch here and we can interconnect you and that's actually what was the first step. So that aspect is very important. If an exchange point is neutral and the people can be trusted I think it will be more likely to succeed.
And the other resource, what's important to see, understand is that most exchange point models, certainly ours, we don't interfere with any of the traffic. We don't go in and look at the traffic. It is up to the ISPs connecting to the exchange point to do what they want with it. Another thing worth clarifying because there has been a lot of talk around whether or not a nation has an exchange point or not. And what's important to keep in mind is that we talk about network topology. First of all, the Internet doesn't act according to national boundaries. But also something that is geographically close or distance doesn't necessarily mean that it is network topology wise closer distance. That's very important. If we are looking at reducing ‑‑ increasing speed, reducing latency you look at where it makes sense network topology wise to put an exchange point. And it is about localizing traffic and reducing transit costs.
Another thing ‑‑ well, two things, and that is that IXP has to keep growing and it is important to have stability of the exchange point to have dedicated staff. Not only would it allow it to grow but it will also provide stable contact points. My other hat to speak that is a root server operator. We operate one of the 13 DNS root name servers in the world and we are looking at Africa among other regions to increase the root server footprint. And we are talking to Michuki and others in Africa. And what we have found and I think that Michuki has given that, told the story about some of the exchange points you put in the infrastructure, but then there is no one there really to maintain it. And if you want to bring an IXP up to that last level, I don't remember if it was 4 or 5 as a thriving IXP with critical Internet critical infrastructures, sorry, you need to have that because that's also where others ‑‑ other value added services will be easily added. So, for example, as a root server operator we are a non‑profit organization and we can find ways of funding the installation of a root server but it is important that we have a stable contact point and technically competent people at the exchange point. So I just wanted to add that and I don't know if any comments or responses to that. Thank you.
>> BERNADETTE LEWIS: Great. Thank you very much for those comments. I am going to ask Akeel.
>> Akeel: Hello. Yeah. I think the issue of exchange point is a critical issue in our region and for Internet development in general. And there is a lot of meetings and a lot of workshops, a lot of discussion about setting up exchange points in our region. But if you look at the map of Africa and you should talk a little bit about that, it has not been evolving that much for the past three years. Why? And I think we may want to look closely at the why it is not making as much progress. Those who already established are pressing but no one is joining. If you look at the map again you will see West Africa, for instance, there is almost nothing there. So is there any political network infrastructure or network topology impact on this? For instance, most of the French speaking countries have a strong monopoly. If it is still not there there is only one operator connected to the fiber and all go through that and for them they have the perception that they don't need an exchange point. They all go through the same.
So how do we present this thing economically to make it a viable? Let's not forget in most of those countries no local content for them to be willing to appear locally. So we have to present this as Michuki said as a long‑term perspective for the development of the Internet and come up with ‑‑ work with the operators and then show them their interest in peering locally and the impact. We have organized the meeting recently where traffic engineering show how traffic engineering can solve a lot of issues related to connectivity cost. How can we really do that without gathering and taking policy ‑‑ big policy decisions because that would take time. Those are very specific points that we need to address because there are many programs here to improve the network in Africa and already talk about root server copy. It is a project that we have run two, three years only and we only install one root server through that program in Africa. And why? Because we want exchange points that are working. So yeah, we can complain all around the world that we want a root server in Africa. Do we interest the network as Internet or Extranet. We connect to the world.
So we need to work more on that and the communities is important. I think that Kenya has made significant progress because the community was strong behind. They work together. They also benefit from the satellite connection at that time where most use satellite. We have to present the exchange points for those countries not taking it may be somewhere where we are not looking at.
>> BERNADETTE LEWIS: Thank you for those excellent points. Again I cannot overstate the need for education and public awareness and speaking to the constituents in a language that they understand. And as I mentioned before the Caribbean experience engaging the politicians at a political level to help in steering that sort of direction. I mean you have spoken to the issue of competition. Is the market liberalized? Those are the questions that you have to ask. And those sort of questions you know you need to raise at the political level. There must be a political will or else those countries that have not benefitted from having an Internet exchange point or a level 4 Internet exchange point will continue to be severely disadvantaged. Can I go online now and see ‑‑ get some comments and some experiences very quickly from our online participants?
>> We have a question from Palestine. First we would like to thank the panelists and organizers and they have a question. What is the best practice for managing and regulating the local exchange point? And they have another question, we need to know from your experience about the regional exchange. Is it better than having only local exchange points? And also they have another question. We need to know the cost savings in percentage and traffic and host.
>> BERNADETTE LEWIS: Michuki, do you want to take the first one, best practices for governance of an IXP?
>> MICHUKI MWANGI: I think there were two questions in that, managing and regulating. So the managing model is a tricky one because I would prefer not to prescribe a management model for the exchange point. The most important thing is for the Stakeholders to come together and agree what model works for them. The default and most acceptable and most successful model that works for Internet exchange point is non‑profits. They tend to go well and gaining confidence of operators. And two being accepted by future members who come to join. It is a non‑profit. They are for the benefit of the community and the Stakeholders that are involved. A non‑profit model tends to win ‑‑ tends to be more acceptable and tends to strive better in the experiences we have seen but that doesn't mean it is the only model. There is a study I tried to do some years back to see what's ‑‑ how the diversity of exchange points in Africa and I realized that we actually do have quite a diverse number of models in terms of where we have those that are run as non‑profits. Those are run by the university and some actually was set up by the Government.
The question is what do the Stakeholders think or will actually appreciate as a model that works for them. That's going to be the fundamental thing because those are the people who will be participating in the exchange point. There needs to be sustainable care of it. If they are not a part of the model then the exchange point will not solve their interests. That's the fundamental thing that needs be considered when looking at the model. Government does that have a role to play. And the experience we have seen with this ‑‑ with especially here in Kenya is that initially the exchange point was for ISPs, but when all Governments go e‑Governments and they need to connect to the exchange point they start asking the question so are we buying services from you or what exactly are we doing with this. Because it is an exchange point, it is voluntary. The model becomes tricky. So the model that you pick should allow for a broad based type of engagement, not just for ISPs but also Governments can have a seat on their ISP board and participate just like every other member. So the model you pick should take those future considerations onboard.
>> BERNADETTE LEWIS: The second part related to the evolution. She is asking about regional IXPs but I think that was explained. It is an evolution. You cannot decide you are starting off, you are going to have this thing. So I think we will just pass ‑‑ you want to comment?
>> I just wanted to emphasize that you cannot start at regional. You are going to a regional exchange point, you can go to London or Amsterdam.
>> The big exchange ‑‑ you can come to Kenya, for example. But that only resolves your issue when you are connecting to content that is available there. You will still need access to the local content. Now that means you cannot ‑‑ essentially a well designed ISP will be connected to more than one exchange point, a national one and a more regional one. The value here will be do you want to grow your national exchange point to be a regional and a global.
>> BERNADETTE LEWIS: Sure.
>> Just for a second to talk about the regulation side of the exchange point, in many regions in Africa if you look at the regional economical commission test they are starting adding exchange point as a need. So the policy side is there but why not picking again? Because those regulators are seen as protecting the monopoly. They are not coming because it is coming from the top to favor the monopoly. It is not only about regulation, but the mindset and about the Internet business itself. How it works really. Why do we need the exchange point and that is also some area where we need to work to educate regulators.
>> Can I add one? So the point is that Kenya is the only IXP I think in the world that has a license by regulation from the regulator. And it was at battle with the regulator to get them to let the exchange point operate. So the point is exchange points should not be licensed. They are not licensable. It is like saying you are licensing the switch you have in your office because exchanging traffic there and exchange point is just a switch. So if you license the switch in an exchange point you are pretty much license ‑‑ requiring a license for every other switch in every other organization. But looking at it from a, you know, and this is where the question of how does the Government get involved. In our case the licensing sort of becomes that role of how the Government gets involved. But I think looking at Multi‑Stakeholder partnership is more of an ideal approach than a licensing approach.
>> BERNADETTE LEWIS: Thank you very much. Wrapping up now but we will take one more intervention from our online community.
>> We have a question from Cameron from Internet Society. They are saying can the Internet Society of Cameroon support an establishment of an IXP in a developing country. Can the Internet Society of Cameroon support technically and financially an establishment of an IXP in a developing country and that question is for Mr. Michuki Mwangi.
>> MICHUKI MWANGI: Okay. So again this goes back to the whole governance issue. It will be very risky in all manner to try and bind the exchange point to an existing organization. It will be a Stakeholders' decision. The Stakeholders will have to come and say do we think that the ISOC chapter in Cameroon is the most ideal location to be the host and the entity to run the exchange point. And the answer to that is they should talk to the community. The ISOC chapter in Cameroon should facilitate the discussion in Cameroon and let the Stakeholders make the decision that maybe the chapter would be the most ideal or they should set up a separate entity. So that's a discussion and that's the 80% of the social engineering that really needs to take place.
>> BERNADETTE LEWIS: All right. Thank you. We have ‑‑ we will take two more very quick interventions. And thereafter I am going to give the panelists about 30 seconds just to make some final comments. So two quick less than a minute interventions please. And identify yourself please.
>> I am Jim from Tanzania. Perhaps I should just quickly say in Tanzania we have established four Internet exchange points because of the size of the country but the challenge ongoing now is how to make them handshake. Internet connection directly connected between the four Internet exchange points. Perhaps my quick colleagues, ISOC or PCH collectively or individually offer taking assistance to us to do that business. That is my quick intervention.
>> BERNADETTE LEWIS: Absolutely. Yes. I am answering on behalf of the Stakeholders. Definitely, yes.
>> My name is Kim We (phonetic) and I come from Nigeria. I know we have exchange point and they tell us to keep local content local but what I don't know is is of benefit to end users. And I note in Nigeria and exchange point that has the backing of the regulator. Thank you.
>> BERNADETTE LEWIS: The benefit to the end user in Nigeria.
>> If I can just jump in, the OECD is undertaking some work with UNESCO and ISOC that we did a joint paper that looked at the development of Internet in countries and we were able to show that if you have a more developed Internet infrastructure also produced more content. We weren't able to determine the causality, which comes first. And once the Internet there is more Internet infrastructure that makes the local content providers provide more content. I think that's an interesting area for the research, but we do see a connection between infrastructure and the amount of local content.
>> BERNADETTE LEWIS: I think that about wraps it up. I am just going to give 30 seconds to each of our panelists just to make some very, very quick interventions and wrap up the points that are most ‑‑ that need to be made at this point.
>> MICHUKI MWANGI: I think in closing, my closing remarks will be that Internet exchange points are catalysts of Internet development in that they ‑‑ there are a lot of other ‑‑ I will say actually Internet exchange point creates what I would like to call an ecosystem around it. Without an exchange point the ecosystem does not exist. The moment you establish an Internet exchange point an Internet eco‑ system starts growing in that you have all these other industries and sectors will hinge themselves on the success of the Internet exchange point. And as a result of the impact is that you have better Internet access and you have better penetration of Internet access and everything else that you are trying to achieve with respect to Internet access will be achieved because you will see that a CcTLD will actually work better if there is an Internet exchange point. Without that it doesn't work really well. So it is really, really important that we look at having an Internet exchange point that will catalyze the growth that we are able to see.
>> TAYLOR REYNOLDS: Just really quickly I would say there needs to be some emphasis on competition and on making these markets more competitive. OECD has been successful in making OECD countries successful. We seen them lasting in parts of Africa. One of the findings of this local content paper is need to promote.
>> BEVIL WOODING: Internet exchange points are a fundamental component in the creation of a healthy, robust Internet. It should be exchanged locally. There is no country jurisdiction that is too small. We have seen that regional cases where small countries such as the British islands and Grenada and Google and ACME are looking to locate them.
>> BERNADETTE LEWIS: Thank you very much. I think that is an excellent wrapup. You have shown the appreciation long before I had a chance to ask you and I want to pay tribute to our online participants. Thank you for joining us and we appreciate the comments. And let's go out and build Internet exchange points and where we have this let's expand them. Thank you very much for participating. Thank you.