 First speaker of the day, day three. We have someone here, whose name I have to pronounce. It seems like some people in Holland and in Belgium call him Swa, but that's another interpretation. But Stefan is not really the German Catholic priest that you would be googling now. No. But he has as well a very special interest in networking security and measurement. Skills that his fellow nameholder that you can find googling on Wikipedia presumably is using as well. Swayo is a well-known player in the internet field here since more than two decades. And being involved in various programs, events and comedies like if I pronounce this right Euro 9 or Euro IX, yeah. And Dinock. So some of you probably know him. He's as well a founding member and managing director of V6 for 15 years. So that makes him besides being an old fart like me, he's above all a time traveler, an everlasting wanderer, an explorer, a learner and heavenly willing to share his experience and knowledge with all of us. So I have the pleasure to introduce to you Swayo. Please. Enjoy. Perfect. So good morning everyone. There are a lot of people here. I have not expected you all here as I was reaching the venue place in the morning. It was empty there when no one in the tram and no one on the streets. And it's good to see that some made it today. Even it stays free and I know that's really hard for you. I really thank you very much that you joined me for this talk. So as I said before and as it was announced, it's a little bit about net neutrality. This is the red line through the whole presentation. It's a big buzzword. I'm not really a legal guy. I'm not an economics. I'm coming from the network side. And don't be disappointed. I will not tell you anything about big legal things. I think there was a talk yesterday by a lawyer, which was much better than I could do it. And so don't be surprised if it's all the network centric. So as I work for a company, standard disclaimer, if you like, read it later. So what's this talk about? I will show the commercial aspects and how they influence the internet and especially the internet access. And this is the only part of the internet where we actually get numbers because everything else is some way in the hidden, in the cloud, in the dust, in the fog, name it. And then I want to give you some glimpse what arguments are used by advocates and opponents of net neutrality. That will be the last slide. And this all was inspired by a movie made a few years ago. And I have their presentation during midnight. All creatures welcome. Maybe some of you have already seen this. I have a small part. Actually, it was a three hour recording session. It was eight minutes left at the end. I'm not as lucky. We have 120 slides. We are now down to 80. So I have to hurry a little bit. So let's start. These are the topics I want to discuss today. I want to present today. Early days. Where it all started. And funny enough, the first paper you can read, which has a little bit of a relation to the internet, is intergalactic computer network. Intergalactic. 1963. And we have then the ARPA net. Most of you maybe already heard of it. This is a predecessor of the internet, as we know it today. Operating for a really long time, 27 years. And it was sometimes misinterpreted as a DARPA net because of the naming. The company, actually the entity behind it was the ARPA. ARPA was later also called DARPA. Therefore, it's DARPA net. It's the same network. No difference. Millnet was split off later from ARPA net. And I think the most important part of the schedule is the NSF net. And then later the NS net. There will be some slides explaining why they are important. So that was a good time, by the way. On one page, the whole internet. You cannot do it today anymore. I think we would need all walls here in this room to have only a tiny glimpse of the network looks today. But this fits on a piece of paper. And actually, if you look at it, this is a quite nice mixture of universities, say, Stanford. I can see their UCLA, USC, Rutgers. The right blue file. There's ARPA. There's the RAND Corporation, which is a commercial one. But there's companies like DEC and Xerox. And as you see, in the beginning, there were already people not only from the research community using this network. There were also commercial companies involved in it, even if they were on the applier for the hardware there. So, as I have said before, NSF net. I think this is one of the important networks they built. National Science Foundation has built five new supercomputer centers in North America. They wanted to connect them. The problem was, then, this researcher at this time has, for every computer, if a supercomputer was built, he has a new terminal on the desk. Think of it. Five supercomputers center, three others. You have eight screens. And every screen has a different command set. That's really boring. And as these people try to do, don't want to do more work than they actually have to do, they're really trying to make it easier for them. So, they said, okay, it would be nice to have one terminal. And that connects to every, these big computer centers. So, NSF net was founded. NSF net actually exists from 81 to 95. 85, sorry, to 95. Bandwidth is funny. It starts with, we would call it ISDN, digital network, 64K bit, 56K bit. This is American speed and went up to 45 meg bit. And later, they were also 600 meg bit. But this was in the last years only. They invented something called a tier. Anybody heard of the name? Tier in the relation to network? Thank you. One. Maybe some guys from the research community in the front. So, what is a tier? This is a definition for this, a level or grade within the hierarchy of an organizational system. Actually, it only says they're the big ones. They connect. They have the eagle view. They connect every location. And then you have more distribution layer where you have two, which maybe you could think of big cities or regions where you have an operator covering this area. And the third, the green, tiny circles, that could be an access provider. Maybe some of you have already heard from something like in Berlin or SNAFU or, yeah, but NetMB Express is all Berlin Internet provider, smaller ones. They would actually be layer three. So, the goal was actually within SNAFU, the goal was research and education. But they phrased it primarily, that does not say that they're not allowed to do commercial things there. And they set up an acceptable use policy. Actually, one of the first policy ever invented for network and that not states that this acceptable use policy is only for research. They said, it's not important who you are. Accept our rules and you can use the network. Nice idea. Everything non-commercial could work on it. And the NSF, National Science Foundation, actually does not monitor it. So, you could actually anything you want to do on it. And that was the reason why they got problems with the regional ISPs at those times. I wouldn't call them ISPs in these times, call them regional networks. And the problem was that, actually, what was the killer application in these times? Anybody knows? World Wide Web was later. UseNet, NetNews, Mail, all these old really great stuff. Some have already heard of UCP and dial-in and even, I think they dropped it a few years ago, this dial-in stuff in Linux distributions and other operating systems. They cancelled it because nobody uses it anymore, but it was the basis. The killer application was email. They have five, six, a dozen different email systems in America. And with this NSFnet, they could connect these mail systems. So, you could send email, if you were at Compuserve, to TimeNet, to the research community, without having five different email accounts and to use your modem to dial into five different platforms. You could use one and it was forwarded. So, as I said, interconnecting mail servers, actually, they were more commercial. Give you some names, Alternate, later called Unit, PSINet, SurfNet. And you see now comes Economics in the early stage. So, Economics covers both commercial and non-commercial aspects of the Internet. And there is really hard to find a really good definition. I think I found one which explains itself. I've read it for you. Internet economics describe various facets and how we build, operate and use the Internet, and look at these activities from a perspective of economics. Nothing more to say about it. RNS. So, what is RNS? This is also one of the last slides for the first part of the presentation. RNSnet was actually founded by Merit IBM MCI in 1991. If you look at the date, it's actually with NSFnet operating to the late 90s. It's actually run, has two hearts in his beating in his breast. There was non-commercial and commercial. The non-commercial part was the NSFnet backbone. This is more the tier one. You have seen these red clouds there in the tier presentation. That was NSFnet backbone. Then you have the more commercial aspect called core, commercial plus research. That was the actual backbone where it runs on. They shared the same media, research and commercial companies. They used the same backbone there. It was later sold and was switched off 2015. Actually, three years ago, actually, this is really not happened decades ago. Only three years. So, these things live a long time on the Internet. So, start of conversation of the Internet, definitely with RNS, definitely with RNS core. Then somebody said, use net news. The first mailing list pops up and asks, actually, who pays for the Internet? That will be my next topic. Who will pay for the Internet? Because you will not get everything for free. Even the research community, even if they have deep pockets, as an example, ARPANET was founded with a million US dollars, which they took from a missile defense system. That was actually the basis of the ARPANET on base Internet views, was money they actually used for defense system for missile rockets. I think it's quite impressive that it was possible. 1 million euros, and this time, it was a million dollars, sorry, was really a lot of money. So, network access points. Network access points, as you see in the hierarchy, you have to hand over traffic to the next level of your hierarchy. It just goes up. And these points where you have the all connecting, where they all connect, they call network access points. And that was a problem with the first network. Actually, you have to use the NAPS. And you have to pay for the NAPS. So, it looks like we have used the tier one presentation, a few slides before. And as you can see, the red clouds, think of it as the NSF backbone or the NS core. And they changed a little bit of money because they were all over North America. And it was not for free. But interesting is the second part, then now look at the blue boxes. And the blue boxes, they have to pay money to the NAP points, to the network access points. And all the traffic, they transfer from 2 to 2 and say the left is west coast, the right is east coast, it's actually, you pay twice for your traffic. Because both parties paid for the traffic to the NAP. In a way, it makes sense because it was a good service who could afford a link between the west and the east coast. And these times it was pretty expensive if you wanted to have a really good bandwidth there. So, but what's the other case? If you're both in the same data center, and maybe the other node, the other blue node is standing in the reach of your hand. Why should it be connected to the NAP? Because it could transfer all the traffic to a NAP, pay for it, and send it back down. Paying for traffic, which I actually could transfer if I put a cable between these two sides. And that was the starting point of the Internet Exchanges. And the Internet Exchanges were something who aggregates this approach to put these connections inside the data center, put them together, and give them the opportunity to, we call it peering, to transfer traffic without paying for it. Both sides could transfer it. That's the ideal approach. So, that was it for the NAPs. So, there was Europe 2. Yeah. We have the DfN. Somebody knows these guys. They have started 1985 with the network based on OSI protocol. I will not talk about OSI that would fill an hour. But it's definitely not compatible with the network we use today and call the Internet. There were two other universities. The older people maybe know them. This is Unido on Xlink based on universities. One is in Dortmund, the other one was in Karlsruhe, which actually have already connectivity to the upper net, to the Internet in the US. And they used their network and it was quite expensive. But that was actually the foundation. Not the DfN. They switched over later. It took them some time to accept that they cannot have the search community put on OSI, even if it's maybe the better protocol. I don't want to discuss it. It's, yeah. So, individual networks and association was founded in 1992. It was quite old as I've seen this. I thought, oh my god, I have never thought that it's so old that they started so early with it. And I have to say, good work guys. That was really good that you were there. And a tiny side note in history, the English guys already had in 1973 connectivity to upper net. 12 years before the German scientific network started with OSI in Germany. So, well, okay. German telecom. Everybody knows bit.com text. I think we all have seen this logo more or less in the last 20 years. A lot of these congresses. And it was one of the first texts of the CCC. And I was related to Internet. They were 850,000 CDs sent out with a Mosaic browser when they used the BDX platform to dial into the Internet. Even it was low speed. That was one of the reasons this was actually the starting point. Dial-in connectivity over the BDX platform. And there were also other companies like CompuServe and ROL. Also, dial-in pools at universities. I can remember the time where we have technical university in Berlin. We've had eight modem lines to the Internet. We're really crowded and we have these, you know, you have to dial with your finger. And it was not a pushing button, so you have really to dial something there. Rotating dialer. So, then they were smaller. I already have said some names, NetMVX, NAVU, other companies. You have to think of them as these tiny green bottles in the tier models who have actually access to the Internet. And give you dial-in platform there. So, other services. Was there a competitor in the market? Yes, there were. Some people maybe remember the name. FEDOnet. Maybox systems. They were quite popular. And after Internet was established, they were all gone. So, nice summary for the first half. Day zero of the Internet when did it start? 83. TCPIP protocol was switched on in the ARPAnet. And the old protocol was switched off. That it was the first time that a network 100% ran on TCPIP. The first email was received and cut through in 84. The e-domains in 86 and in 89. The hypertext protocol was invented by a guy at CERN. And I personally think this was for the commercial Internet and for the Internet for the private customers. This is an important thing because that was the first time that this cryptic Internet suddenly got a face. Suddenly got an easy to use interface. And in 95, already said, before German Telecom sent out these CDs and people started to use this stuff on their PCs and they had the first experience of things. Even if it was really slow and it was maybe not the best browser in the world. It was a browser and you could see pictures, you could see ceilings and so on and so on. There's a whole story behind it. I would unfold it. We would have breakfast tomorrow morning. It's really, really a lot of stuff happening then in the meantime. Therefore cut here. Maybe this is part of another presentation should be done by any other guy on me. We'll see. So network explained that this maybe for some of you very boring part as you already know or maybe work in this area. But let's look at the traffic flow on the Internet. And this is a very, very, very, very simple model there. And don't blame me for it. If you think that it's not right, we can discuss later. So starting in the middle, the core. And the core, we have now again the tier model we have seen before. And the tier model also the links between the big providers there which actually do not pay for them for the traffic. As you can see, they're a good dozen companies today. This is an actual list from this year. Quite an additional lot of companies from the first world. You see no South American or African companies there. Also not too many Asian. But this is actually the people who run the backbone between the continents. If you think of the clouds maybe as the continents and the cross links between they will supply you with this. So then we have the tier 2. You pay for this. Some money if you want to have traffic there and exchange traffic. And if you're in the green area in the tier 3, you pay a little bit more to get access to the Internet. And what I do not show is if it would go a little bit further down, then this is the access. That's you actually using the Internet at the end. So in an ideal world, traffic would be for free. As you can see, between the green bubbles there, if they would have a direct connection, who should pay for this? It could be quite cheap there, but it's not a reality. Reality is in worst case that you have to turn traffic from tier 3 to tier 2 to tier 1 and then back again to tier 2 and tier 3. So you pay a lot of money. Both sides would pay for this. And that was the idea in the beginning with the Internet, the NSF net, like the American sort that could work. Reality is more like this. You have this tier 2, which is really the backbone of the Internet, I would say. And there you have links between all these providers at this level. And if you see some of them have zero dollars there, that could be an Internet exchange as an example or a direct cross-connect between two companies in the same building. And now it gets cheaper for the green guys and actually also cheaper for the guys sitting behind the green bubbles. And it could even be better if they have a direct link. And as the interconnection prices get cheaper and the long distance links getting cheaper, this is also happening today. Good. Now access. This is a really complex sheet. It's actually four kinds of devices. By the way, this company, I use the icons for the CPE devices there. They have the best icon set. It's there are 100 other companies who consume supply with these boxes. So what we have there starting in the lowest is the 4G, maybe in the next year was 28 GHz, the 5G networks in the metropolitan areas, the not bound on wire stuff. And then you have the fiber to the home that's, you get the fiber there and some magic technology transports lights into your flat and then there's internet on it. And some other stuff we will see later. Then the classical CATV networks, which are using coax cables. Now they have bandwidth reserves up to 2Gbps and now sell 500Mbps up to 1Gbps and then going up the orange ones is the copper lines. Let's see. Two wires of copper going into a flat once used for telephone. Now it's actually all internet protocol. Yeah, it's actually transporting internet over another protocol. But there you have the distanciation. If you look at the first gray bar on the left and then the second is more to the right, this space in between you call the last mile. The last mile is actually from the axis where you get access to the internet to the core to your own flat where you're living, where you have your termination device, your customer premise equipment, CPE. So don't forget, this is not access. There's another infrastructure between core and the last mile. It's the access layer. You may have to distribute because these areas you have to cover are really big. So you have another layer of network stuff and equipment in between. It's called the access layer. This stuff is connected by fibers. Later we will focus on the DSL links most common. I think they are the only real numbers we can compare. Yeah, coming through the content. Content delivery. In the middle it's a core. It's a tiny world map inside. Then you have all these areas. South America, South America, sorry, Europe, China, and then you have Australia and Africa. Africa will be later inside. Sorry, forgot it in this slide. But it's there. And there you have these big cloud providers. Everybody heard of AWS, Azure, Oracle, and even companies like Salesforce and SAP who actually sell only one product are now put into these cloud operators. They have data centers with a lot of servers inside which are located in your region. There are also initiatives like in China. I hope I have pronounced it right. We have also cloud services in Germany which are really focusing on the local market. The market grew. They are more competitive on the market nowadays. Their market is changing fast. As you can see the blue arrows I have there, this is actually Internet protocol. That is something where the traffic you use is transported. If you take Google with Alphabet, if you have to Google search, you are actually running on these blue links. But nowadays they also connect these areas with their own private infrastructure. Traffic comes closer to you and you only see, and this is a bad thing for research, you cannot see where the traffic actually comes from. Giving an example, I think we have some time for this. Think of it the way, you have a service on your smart device which has a JSON interface which only pulls data and presents it in the nice way on your device. And then you connect to Frankfurt to AWS. This is the only example you can go to is the same story. And then where is the data that you actually store? Is this in Frankfurt? Is the older personal data stored in Frankfurt? You do not know. Because the same data could be in North America, it could be in Asia or it could be split in both locations. And all the data from this point where you actually pull the data to your nice interface, all these data, what happens behind, is transported over their infrastructure. Maybe based on IP and Internet protocol, but it is not public Internet. You cannot see what kind of traffic flows there. And therefore they build actually a second network layer there which is a quite interesting thing, but could be think of the tier model, which traffic is actually going away, actually the traffic from the tier ones. Because the tier ones normally transport traffic from continent to continent, but if this is actually done by the cloud providers on their own, with their own equipment, with their own links, there is no traffic on these links anymore. When you give me another idea, you have these transatlantic fiber links, undersea cables, more than 90% of the traffic is not Internet. This is enterprises, this is commercial traffic. So the Internet itself use is actually a tiny part of the traffic produced worldwide. There is a good reason why. So, content, classical one. We started with a server somewhere, have a really thick pipe there, and then we connect to the core and then it is distributed worldwide and then you get complaints from Asia that your server is too slow and what you can actually do, you can put in every location a server, it is quite expensive and then somebody had a really brilliant idea and said build a content distribution network. There are some vendors, names are on the right side of the sketch, who built caches these red tons I have there. They put all of these caches into the core networks as near as possible to the client to make it as fast as possible. These caches are normally transporting your content. By copying this content to these sites and by DNS redirection, name service redirection, you are getting the traffic from the site, which is nearest, which means that you get it fast and you do not have to wait a lot. There is later a sketch that explains why this is important, why you have to transfer traffic really fast in this network for economic reasons. Then there is a special case, audio and video, first look it looks like CDN, it is actually the same thing if it is a text or it is a movie or an audio file, why should I make a difference there? Actually there is a difference because you distribute these audio and video stuff because of digital rights management, things and issues. There are different areas, there is different content, there are even services that you cannot get, one of the sad things maybe as an example in Europe BBC has services online which you cannot access from central Europe. That is not the cause because they have Brexit that was well before, so the problem here is that these content is most of the time local content. Now maybe some of you heard of it, OTT and I did not miss an O there is not called OTT, it is over the top. Over the top means these are operators who do not have their own network, they do not have some access network like Time Warner and US who owns the cable network and owns film studios, so they have a line of producing content, storing it, transferring it to the end customers and get paid by the end customers for this content. Now these people do not have this, they really rely on the stuff that you have running network to the customer which has enough bandwidth and enough quality of service that the experience for the end customers is a good one, a great one, because OTT wouldn't pay for it. And this is only a definition for this who is actually doing the search engines, social media platforms, all of Rydion demand, there was an example before in the sketch and the last topic is I think this thing where all the discussions about economics and who pays for the internet actually starts especially in Germany, carrier does not participate in the earnings and the earnings here can be do different kinds of earnings. The one is cash if you have a connection to the internet you pay cash money these OTTs can also be paid in a different kind of money, your personal data, by advertisement as an example, you're accepting this and this is the second that is much stronger and this advertisement stuff, no access provider sees any money from this, this is actually really privately owned by the OTTs and I think for the OTTs they do not care about the cash you pay, most of them they want your personal data, has much more value than every bitcoin in the world. So we now have a nice model we can make a shell with access and the next shell is the access layer and then we have the center with the core and then inside these middle part we have cloud providers, CDNs and other services, actually the stuff that people and access wants to have and is it really true? Yes, it is true, this is a sketch from KIDA they have really nice pictures of the internet there, if you're interested have a look at it, there's a link there KIDA.org and this is a map from 2017 and you can really see how these onion shells develop there you have the outer layer with the access and then it goes inside and gets more dense and more dense and dense and then you see there's the core, there's actually really a core of a few sites which supply the internet with content so internet economics, mentioned it a few times before, now we are diving into this topic and in this topic we have the access, everybody has seen the street cabinets in Germany, it's called Carfautsit really nice name for Kabelverteiler and here this is regulated by the Bundesnetzagentour German regulator and they said everybody should get access there, therefore we have these frames has to be opened for other carriers too and there are actually two points where you can use the kappa cable this is the Hauptverteiler, this is the master distribution frame this is a building which is about 1500 meters 1.5 km away from you, if you take the length of the cable, that was the classic one, this is the place where you got 10-15 years ago your iStand connectivity, your internet connection this is also on these 1500 meters you can run 80 SL with 16 m bit, but nowadays most of these people have 50 m bit or 100 m bit or even more to do this, because if you have more bandwidth you have to get closer to the customer so therefore you have these Carfautsits, these street cabinets where you have active equipment inside and last discussion we had the bit stream access and the bit stream access this is the stuff where actually it gets more important as a vectoring camera so 100 m bit links to the access has happened the kappa bundles you have there can only be run by one operator, so you cannot get the physical cable anymore, you can only get the IP traffic on it and you transport it in layer 2, layer 2 is actually only a transport layer you have to put layer 3 or not, layer 3 is actually the IP, the internet layer so you get a layer 2 connection there and this is regulated and there are prices for this, this is public stuff you can read it, but in this Netsargentour, you have to look a little bit, but you will find it there are lists and this is all public, you can read this stuff how much money cost it, if you go from master distribution frame, this is a thing which is 1,500 m away from you, you pay 10 euros per month, by the way these prices are all without Mewatsteuer, so don't be surprised that you can have on your bill different numbers later, this is 19% less I cannot compare with these prices are always without Fed, so as you can see on the right side, this is from Buga the price actually went really slowly down from the price, so it's actually so, and if you compare it with the price of the internet, if this is the connection to the link, the price of the link as described there, the internet price goes down this way, so actually price for the link, if you do a couple lines, stays the same, and the internet goes 100 times cheaper, the traffic you transfer there, and in the same time frame, so physical stuff, yeah, it's a lot of money inside these things so from the street cabinet, which is 300 meters away, a little bit less, 100 meters then it's only 6 euros, 7, 7, it's a little bit less, but still enough so as I mentioned, important keywords here is vectoring and second important keyword layer 2 access, you have to go to a point of interconnect from the operator of these vectoring devices, and then you have to connect your equipment there, and then you can hand over there of the traffic of your customers yeah, and how does the carrier pay for it there's nothing like a flat rate there where you pay a base price and there's some little bit of traffic included but if you want to have more quality of service traffic then you have the problem that the you want to have more, then you have to pay more for this as an example how this is described, there's later a sketch for this here this is how it's described, it includes a monthly total traffic of 1.5 megabit in average, and a real-time share of 0. and then you have stream applications, so they split this traffic up and as you can already see, they already look into the traffic in the bit stream access and there is quality of service involved this is actually the price for the internet if you take the prices from two vendors, the Stroman Telecom and O2 only important here is the orange part this is the price actually of the internet traffic you have on these lengths, and as you can see, this is nothing compared to the rest of the price, the blue one is the hidden price, this is all the stuff from access to backbone and everything which is there included so, the provider sells you traffic for a gig bit of 60 euros and if you look at the market, low quality is about 200, high quality 500 euros for a gig a bit of traffic so, it's actually three times more expensive devices on the free market up to 42, we call this an over-baking factor, this is one of the trade secrets nobody will tell you, because all the business cases of the carriers depend on it so, internet having core defined that there is a huge difference in price as you have seen in quality, and there are different important things here you have to think of it has to be redundant and resilient so, normally you have two lines, private have only one link you have two links there to get the connectivity there safe, so, content delivery in former times, every service was transported on a different layer in a different way on the net, and then they put IP there, IP is not internet, it's a protocol IP, it's not the internet and what they did, they used this to migrate all the stuff to IP to make it cheaper, but they still think it's a different thing here, and then the auto ET came, and they have the same services but they use the part of the internet, you supply to your customers so, you have now two VOD services on your network, one is in the eyes of the access provider, their own, and the other thing is the VDD service, which is run on the internet they need bandwidth 4, and who pays for it, I give you an idea call of duty 100 GB of download, that's equal to 14 hours of 4K video, each video has 7 GB of traffic so, why do you care about it? fast internet, normally means that people who have a slow connectivity normally wait up to 30 seconds before they say 50% of the CPU is low, content is not important, I mean drop it, and if you have a fast connection, this time shortens to 5 seconds, so if you don't want to lose customers, you really have to have fast connectivity there so this is already said during the talk, we can skip this one as I'm a little bit running out of time, I want to go to the net neutrality one of the last points, net neutrality defined by the US in 2012, 11 the principle is that internet service treat all data on the net equally and not discriminate or charge different by user content, website, platform, application type of attached equipment or method of communication, a lot of topics covered there this is the, I would say this is based on the services of the old internet not really taking into account that there are OTTs in the world, so what we have in Germany is the regulation and the regulation covers a lot of things, one of these things is net neutrality so topics discussed in the public is how neutral is the traffic is traffic equal, shiny butter, bill debate, deep packet inspection, you will see this in the press on a regular base that these things pop up there and last thing is two classes internet, this is one of the strongest arguments for people who say we need a regulation there, this is a list of pro and contra arguments, I think only to say is the last line I think is in a way the importance, it's a good summary of all the topics before, the people who say we need net neutrality say nobody wants the internet by design, remember as I started the presentation the internet was equal for everyone, it was shared by commercial and research so by design actually there is no quality of service inside that was not differentiated and the contra people say they own the network they can do what they want with it because they have paid for it so it's their property and then this can happen, this is a Portuguese provider and this is really scary in a way because they have these smart net contracts where you get flat rates for bundles of services what I really find scaring is this, if you look in the middle on the last line there is even an email service included there which was really surprising for me, why is Google mail there I can think of video on demand services, yes, no problem but why I have a mail service and this is really a scary thing if they start to put quality of service on the base services and not only on the video on demand service as an example I would say I would be scared so Outlook, that was requested by the, and I think I am nearly at the end this is an Outlook, this is a mainstream text, you can read it in the paper so what you can see here, there is growth rate between 15 to 58% depending on the services, video is one of the strongest with 58% and there are numbers I do not know where the source is and I think we have to really look at this, supply to you this number, Netflix doing 15% of the internet worldwide, that is a huge number and that more than 50% of the internet traffic we have today is encrypted, which is a good thing by the way I like the idea that a lot of these traffic cannot be put, cannot look into the package with deep packet inspection, can say okay this is this traffic or what kind of traffic is there transported I forbid Google, 40% in the APIC region, we will see this in Europe too, less I think, because we have also strong competitors here, and what are the challenges what we do actually here net neutrality will be redefined, that means we have to take into account that there are these OTTs, they exist, but we have to find a way to handle it, I do not know the answer to this there has to be a discussion, which we maybe go on for a few years and the risk we have here is the wallet garden wallet garden is everybody has a mobile phone and what you have there is that all the services you have in a way is in the garden, which has a huge wall around it and these wall has gates and everybody who wants to go through these gates to you as a customer has to pay the one that is the gate, this is a wallet garden, it can be a good thing if you have a golden cage, then it's okay, maybe you are lucky but there are a lot of examples where this wallet garden is a bad thing, because it limits the access and this wallet gardens in the mobile world could now swap over to the access world, and if this happens you have seen this Portuguese provider, where even a mail service gets something special, you have to pay 4.99 for it, this is called wallet garden, do you want to have this? I wouldn't like it more encryption, more internet protocols will be transported in tunnels who wants to look for the keywords DNS of HTTPS and quick more web browser based applications is ongoing more and more stuff, you do not have applications on your laptops, you normally do only a good browser to run these things and as I have seen presented before, these links between these different cloud zones will grow, so actually the internet traffic will only happen between you and the content and everything behind it will be handled by the cloud operators and more question mark, paid QoS, quality of service we will see, so the end, personal statement from my side the discussion about net neutrality is needed, we built the internet as a network for exchange of ideas and nearly free access to information worldwide, but this covers only the first world, this discussion, so we shall never forget that it's our responsibility to preserve the original idea of the internet and close refreshing memories, it was enjoyable to do this presentation as I have also found some stuff I have known before, and I thank you all very much for your patience and thank you that I could present it to you Thank you Swio, fantastic I personally have quite some questions here, but I presume there are some in the audience I see someone at number 2, please shoot So the short one is about slide, around slide 60 you showed some QoS numbers around 60, I'm not a country member exactly so where did you get those numbers from, I mean 0.001 megabyte per sec for critical services or something like that are those real numbers, or where are they from how did you come up with them? I repeat your question, where is my source of information for all these numbers I presented here with the traffic and how do I come to such load numbers like a euro for these and the base of these numbers, it's hard to get this I used, and I have a calculation here, but this fork was only 45 minutes so I couldn't put it into the slide deck it actually uses the access numbers and extract this and then I mix it up with the stuff from the mainstream facts later where you see 12% is Netflix and 58% is this traffic and 40% is alphabet, so I put it in and I definitely have to say that it's in a crystal ball this is a wild guess, even if I would know the numbers these stuff is top secret, it's business secret even if I would know them I couldn't tell you and one other one actually that you say that vault gardens would be great, I tend to potentially tend to read, no? No, no, no, you got it wrong, I said you have a golden cage, your provider gives you a lot of things and you are happy not for my view as a network operator, more for me, for a customer you see, you don't have to forget, you are maybe 2% of these people who use the internet they are 98% outside, think of your parents, think of your grandparents these are the normal internet users and therefore I said these people could be happy with the wallet garden because they feel like in a golden cage there is a good support, we would say it's limiting us but we are a minority here. What my main question is is that I sometimes feel that there should be huge technological changes down in the backbone IP, is that a real good thing forward and there are some very interesting ideas like name data networks where you actually address the data that you want to reach instead of the end points, how do you see these picking up in the next decades, do you see any appearance of these in Yeah, so I hope everybody has got this question I see what you mentioned at the beginning in the interconnection of the cloud providers because they are free, they can use their own preparatory standards and what's really working will swap over, it's like this quick as Google invented it and I'm not sure which version they are now but they started it internally and then they rolled it out and then there will be an RFC and then there will be standardization, standardization is not a deen norm, it's actually a request for comment and if everybody agrees then everybody will be accepting it and implementing it sorry, if there's more questions I will be here, I'm happy to discuss this with you but I think there are some more people, there are some more people and certainly if we discuss the golden cage there, think about the new developed world and there the cages are maybe not that golden but they need the environment, I would like to have a sarcasm shield here but there was no time, number one, please I think one of the lessons of the Egyptian spring is that you cannot shut down the internet if you want to suppress public digital if the industry is dependent on it so do you see any moves of the internet industry to kind of create a public network that can be shut down while preserving business to business traffic I would say very strong yes, I see that there are people pushing this and say we built a second, a third, a fourth, a fifth network and parallel to the existing internet, so remember the phrase two class internet, maybe extended it, say there will be an end class internet, one for business, one for paying customers one who pay for the extra quality of service, the third and the fourth and the fifth scenario of this base internet connectivity where you can have not even see a Netflix movie because it's even this progressive download, it takes a long time for this so it could happen and I think this is one of the things we have to take care of that this thing is not happening in the future it feels a little bit like taking first class in Deutsche Bahn from Frankfurt to Leipzig to me then certainly, do you think it's a good idea? Yeah, you end up in a different situation, it's not like fast fast but it seems fast, Mike too please Hi, my question was regarding the revenue streams for ISPs and if it's considered that the infrastructure is acquired with public investment most of the time especially for the big ones which used to be part of the state and used to own the network in the first place I think it's your point that I try to rephrase it a little bit here and maybe you can say yes or no if I get it right you say that if you look at the access, you have these nice sketches with the links, coax and these all these infrastructure was built by companies owned by the government and now they're in private property and now they sell back what we paid with our tax money and I only copy the argumentation we have seen five, six years ago and yes, I think it's problematic in a way and I can say that the part of the infrastructure itself and running the infrastructure, you have to look at it the way that if you want this infrastructure available for more than 80% of the time you have to put money into it, this money has to be spent and it was spent by the carriers in the last ten years, so maybe you could say or maybe it's hard to answer, I would say it's already paid off in the time before it was privatized and the privatized people and the privatization and the rebuild of the infrastructure and extending the infrastructure is done by private companies or partly public companies and there's a good reason why they're allowed to sell this to us because with the existing infrastructure ten years ago it would have been really hard and if you've seen there were no street cabinets with active technology inside, it was a huge investment there, that wouldn't be possible so you're right, ten years ago I would have totally agreed with you that it's unfair, nowadays there's so much money invested into the expo, lots of these companies so maybe they have the right to sell it to us, maybe but the question here is with or without regulation Good, food for thought, please sir, shoot I have a question about over the top services and this is almost a political term I think which tries to say that this is a trick they are doing and I think that I pay for my internet access and therefore I should, after all I have paid for it I should have access to it and maybe it's only the way of the payment if I pay for a flat rate and maybe you have a problem with this over the top services but maybe we should pay a fair price per gigabyte or terabyte of traffic as an end user so why not increase the price of the internet access if I won't have free access to all these services A little more complex answer here the first thing is, video streaming started with two ambits per second we are now at average with Netflix streamed four ambits and with 4k we are going up to ten ambit per second and more problem here is that the OTTs use the network in a way which is not intended to, the protocol itself is normally said that you are an average traffic, so you guarantee an average kind of traffic and what we do now is that we have these peak hours, so the over top providers if you come home at six or five and you switch on your video on demand box you are not the only person doing it, so you have a lot of capacity to build up for this and suddenly at ten in the evening, wherever it goes to bed, not maybe us but a lot of other people, the traffic goes near to zero so this peak traffic, the problem now is here, you can see the movies the banter is there, but it takes you a long time to get it started or it takes in progressive download more than 20 seconds or more than a minute to start streaming and as you have seen in the last sketch if you have a broadband access in people, switch the stream of 50% of the people in five seconds, so what the over top carrier wants from the carriers is that they supply them with the bandwidth they need that the customer is not switching off the stream in five seconds during six o'clock in the afternoon and ten in the afternoon and the internet was not built this way sorry, this peaky traffic is really hard to accomplish and therefore the OTTs, it would be nice if everything would happen, you switch on every full hour and maybe you switch on every half hour, the internet would be great but both of you will switch on it at the same time, maybe at 8.15 after the target show I was paying for this bandwidth and now you say I cannot exit this because all the customers do it but all the customers paid for it actually and this is something which would be actually the presentation, sorry the thing is you are right, you have a guarantee with the bandwidth it's a good point, yes you have this but we are talking about peak traffic and there's a difference between peak traffic and average bandwidth I need a piece of paper and pen but we do not have here but I'm happy to talk to you later about it I think it's important as well that in this universe and outside of this universe there are some questions as well coming from the internet, is there really an important one because we need to close here, you're available after this talk but can I have a question from the internet, please there's no question from the internet, I was here still deep sleeping oh that's good, they're sleeping then you are awake, I take one more question please, yes is there any role for public infrastructure like if local or the state governments took over the last mile stuff, does that actually solve the problem no, we will not solve so give him a mic, maybe here's the answer for this where is the answer, here is the answer, please I'm a small town ISP guy, so we cooperate with a lot of city carriers, small ones in Stuttgart and Trübingen and Reutlingen, so the point is if you look at the cost structure of building the local infrastructure there's a very good point in doing it as a public infrastructure because all these city carriers, basically Stadtwerke dig up the streets of your city every 30 years so there's a business case to build this common infrastructure in one run and not dig it up every three years for water, for power, for gas and things like that so the point is that the economic side of the discussion went into regulating this market in a way that leads to a brain dead implementation where we already dig up the streets every few years for a different part of the infrastructure and we say we do it to separate the different parts of the infrastructure to keep this in a regulatory structure that is given by the Bundesnetzagentur, but it's not a useful structure, but the point is if you try to publicly and politically challenge this structure of the market it's very difficult because all people say this can't be done. So I really invite you the next time to have a talk about this, Stadtwerke the now is small for the whole of Germany and I'm aware of these approaches there, even in Berlin there were approaches like in Amsterdam that some of the infrastructure is public, but it got stuck. Okay, that's the topic then for the C3 or the camp. Yes, sure. Thank you all for being here this early. Please thank to Swio for having us inside.