WEBVTT 00:01.720 --> 00:05.260 Okay, welcome to the second lecture on algorithms for Internet 00:05.260 --> 00:05.800 applications. 00:06.080 --> 00:10.680 Last week I showed you several things on introductory material. 00:11.220 --> 00:17.140 I showed you this, you remember I showed you the app for cooperation, 00:17.600 --> 00:20.740 so for hand-waving and so on. 00:20.860 --> 00:24.260 So if you are interested in downloading that, if somebody is here who 00:24.260 --> 00:29.900 would like to download that, you can do that from that QR diagram. 00:30.380 --> 00:32.340 Somebody here who needs that at the moment? 00:32.920 --> 00:33.140 Nobody? 00:33.880 --> 00:37.800 Yes, you need that, then you can download it. 00:38.680 --> 00:39.680 Do you have it? 00:40.520 --> 00:40.740 Okay. 00:41.620 --> 00:47.380 So I showed you or told you several things about this course. 00:47.380 --> 01:01.000 I mentioned that you should register for the tutorials. 01:01.440 --> 01:02.500 Somewhere I had that. 01:03.180 --> 01:03.520 Here. 01:04.140 --> 01:09.560 I hope you all did that because the deadline for registering for the 01:09.560 --> 01:10.960 tutorials ended yesterday. 01:11.860 --> 01:15.180 So if somebody is here who has not registered, then you should really 01:15.180 --> 01:19.020 hurry up and send a message to Marc Mülthin or Fabian Rigolle. 01:19.880 --> 01:22.580 Okay, that was the introductory chapter. 01:22.720 --> 01:31.680 Then I showed you a little bit about motivation for why we actually 01:31.680 --> 01:34.140 look at Internet and applications. 01:34.720 --> 01:39.960 I made several remarks on the way we actually perceive the Internet as 01:39.960 --> 01:41.240 in different facets. 01:42.740 --> 01:46.820 We noticed the range of questions that one could ask with respect to 01:46.820 --> 01:49.440 using the Internet in different ways. 01:50.060 --> 01:57.100 And the final thing was that we have to look at appropriate algorithms 01:57.100 --> 02:00.160 for using all the information that is available on the Internet. 02:00.560 --> 02:03.180 And we will do that today at more depth. 02:04.520 --> 02:05.920 And let me just briefly see what I have. 02:06.080 --> 02:07.000 Yes, that's okay. 02:07.820 --> 02:12.900 And what I would like to do today, just before we actually start the 02:12.900 --> 02:21.680 course, I would like to briefly make a commercial on inviting you to a 02:21.680 --> 02:28.000 student afternoon, just some event at the Research Center for 02:28.000 --> 02:31.100 Information Technology next week, Wednesday afternoon. 02:31.100 --> 02:37.020 There is an afternoon for students where you can look at what we are 02:37.020 --> 02:39.980 actually doing in the Research Center for Information Technology, not 02:39.980 --> 02:44.300 that far from here, just on the Heide-Neustraße, close to the 02:44.300 --> 02:51.860 Technologiefabrik, technology production site or whatever, or 02:51.860 --> 02:53.260 technology factory. 02:54.020 --> 02:57.840 And here in the Research Center for Information Technology we have a 02:57.840 --> 03:02.940 house of living labs where we work on or where we show certain things 03:02.940 --> 03:05.140 that we are doing there in the Research Center for Information 03:05.140 --> 03:05.560 Technology. 03:06.040 --> 03:09.040 In particular, we are having there several living labs where we 03:09.040 --> 03:15.540 demonstrate our current topics for research and development, which we 03:15.540 --> 03:17.340 very often do in cooperation with companies. 03:17.880 --> 03:21.980 And in particular, there are many possibilities for activities of 03:21.980 --> 03:30.860 students, either as a research assistant or you can write your thesis 03:30.860 --> 03:37.060 there and do other things, participate in the project that we have 03:37.060 --> 03:37.320 there. 03:37.720 --> 03:42.720 And the topics that we address there are service robotics, mobile IT, 03:43.740 --> 03:52.000 satellite navigation, automotive, where the technology for the 03:52.000 --> 03:57.180 automotive area, like technology for letting cars move, is in. 03:57.780 --> 04:01.540 Then industrial automation, how can we actually improve the efficiency 04:01.540 --> 04:07.000 of industry production sites. 04:07.340 --> 04:11.900 And then we have the smart home with ambient assisted living and also 04:11.900 --> 04:14.180 a smart home with respect to energy issues. 04:14.180 --> 04:19.440 So we have a living lab energy management, which is looking at the 04:19.440 --> 04:23.880 complete building, the energy system there, and developing energy 04:23.880 --> 04:27.460 management facilities in order to prepare for a world where it's 04:27.460 --> 04:31.780 necessary to have more flexible energy consumption. 04:32.600 --> 04:36.000 And so far, there's also an office environment where we have certain 04:36.000 --> 04:37.780 technologies in there for home automation. 04:37.780 --> 04:42.940 So lots of things where you can actually do or perform certain tasks 04:42.940 --> 04:45.500 that you are interested in, or just have a look at what we are doing 04:45.500 --> 04:51.440 there and then you can decide whether you would like to get involved 04:51.440 --> 04:51.780 there. 04:51.960 --> 04:58.440 So that was just a short commercial, and now we come to the topic of 04:58.440 --> 05:03.480 today's lecture, Internet History and Technology. 05:03.480 --> 05:08.780 So I will make a brief, or give you a brief overview of how the 05:08.780 --> 05:10.080 Internet actually evolved. 05:11.780 --> 05:15.520 I always get some remarks by my assistants. 05:15.980 --> 05:20.640 You should shorten that because it's, you can just read that. 05:20.700 --> 05:22.700 I don't have to spend that much time on it. 05:22.920 --> 05:28.060 So let me just briefly tell you what the Internet actually is. 05:28.060 --> 05:32.340 We looked at that last week already, so you remember this definition. 05:32.960 --> 05:37.000 The three different, or three major aspects that make up the Internet, 05:37.260 --> 05:41.440 the globally unique aerospace, the protocols, the high-level services. 05:42.020 --> 05:46.400 And now we, in a moment, we will look at the protocols and what's 05:46.400 --> 05:46.980 behind them. 05:46.980 --> 05:51.140 And just briefly, before we do that, I would like to just give you 05:51.140 --> 05:59.500 some information or just look a bit more at the history of the 05:59.500 --> 06:00.680 Internet and how it actually evolved. 06:00.780 --> 06:04.260 Because this is quite interesting to see how actually events and 06:04.260 --> 06:07.980 certain things influence technological developments. 06:07.980 --> 06:14.720 So the real reason for why the Internet was established was that the 06:14.720 --> 06:17.620 Russians started their first satellite, the Sputnik. 06:19.280 --> 06:24.740 And at that moment, the United States got afraid of attacks from outer 06:24.740 --> 06:25.200 space. 06:25.660 --> 06:29.720 And so they established the Advanced Research Projects Agency because 06:29.720 --> 06:34.340 they said they needed more sophisticated technology. 06:34.340 --> 06:41.200 And in particular, they wanted to get a network infrastructure that 06:41.200 --> 06:46.320 would allow them to actually survive an attack on some of the 06:46.320 --> 06:47.940 essential communication nodes. 06:48.120 --> 06:52.760 So that was the starting point for having packet-switching networks. 06:53.780 --> 06:57.420 And so it took a few years before they actually came up with that. 06:57.420 --> 07:03.260 And in 1968, so you see six years after they actually developed that 07:03.260 --> 07:05.200 first... 07:05.200 --> 07:09.520 well, the first ideas about this packet-switching network, they 07:09.520 --> 07:11.740 decided to build a network on that. 07:11.900 --> 07:16.980 It started with four universities being connected to that, mainly in 07:16.980 --> 07:20.320 California, also University of Utah. 07:20.320 --> 07:22.300 So it was a very small network. 07:22.760 --> 07:30.720 And then it grew quite significantly over the following years. 07:31.120 --> 07:35.420 So in 1972, there were already 23 hosts in that network. 07:35.620 --> 07:38.820 It was actually the first email program 40 years ago. 07:39.260 --> 07:45.100 72 BBN is a company which is doing most of its work in connection with 07:45.100 --> 07:47.600 the Department of Defense of the States. 07:49.580 --> 07:54.200 And so here, initially, they didn't use TCPIP because it wasn't there. 07:54.540 --> 07:57.380 They used something which was called the Network Control Protocol. 07:58.200 --> 08:04.200 And then a year later, Vinton Cerf, the current president of ACM, he 08:04.200 --> 08:10.940 designed, together with Bob Kahn, the Network Protocol TCPIP, in order 08:10.940 --> 08:14.440 to initially to communicate between different computer networks, so 08:14.440 --> 08:15.700 different local computer networks. 08:15.700 --> 08:16.080 Yes. 08:16.640 --> 08:17.200 Yes. 08:18.920 --> 08:23.120 Backbones is the... like if you have here... 08:23.120 --> 08:24.760 Oops, let me just draw something. 08:25.800 --> 08:31.000 You have here some host and you have here a certain network where you 08:31.000 --> 08:37.500 have lots of stations connected to a computer. 08:37.880 --> 08:42.060 You might have another site where you would like to also have certain 08:42.060 --> 08:42.640 computers. 08:42.640 --> 08:47.940 You would like to connect those, and for that you need some powerful 08:47.940 --> 08:53.040 backbone, some communication line, which has high bandwidth in order 08:53.040 --> 08:55.820 to support the communication between different sites. 08:55.900 --> 08:59.000 So it's not the communication, like not the local area communication 08:59.000 --> 09:03.320 from the server to certain stations, but it is the communication 09:03.320 --> 09:07.940 between regionally separated sites. 09:08.980 --> 09:13.980 So it shows 50 kilobit per second was the backbone capacity at that 09:13.980 --> 09:14.360 time. 09:14.840 --> 09:18.600 High speed, high bandwidth at that time, and certainly we know that 09:18.600 --> 09:25.100 nowadays we are quite far from that, much more bandwidth. 09:25.100 --> 09:32.720 Okay, so then in 74, so 38 years ago, the term internet was used the 09:32.720 --> 09:38.060 first time by Richard Burke and Bob Kahn in their paper on TCP. 09:38.760 --> 09:41.560 So they actually started that name. 09:42.520 --> 09:47.640 And then there were several other networks designed the following 09:47.640 --> 09:48.200 years. 09:48.680 --> 09:50.600 I will go through that quite quickly. 09:51.360 --> 09:57.580 So in 76, you see, several years after it actually was designed, the 09:57.580 --> 10:02.340 Department of Defense decided to actually use it in the ARPANET. 10:03.320 --> 10:10.300 And then initially it was a network that was built or constructed, 10:10.680 --> 10:14.260 established, because the Department of Defense of the United States 10:14.260 --> 10:17.360 decided they needed such a network for military reasons. 10:17.360 --> 10:24.020 And so then some other people, certainly like, this was designed by 10:24.020 --> 10:27.280 people in academia, and they said, we need our own network. 10:27.500 --> 10:32.460 So they set up a Usenet network, was a decentralized news group 10:32.460 --> 10:32.860 network. 10:33.420 --> 10:38.180 And they also set up a different network, a Bitnet, because it's time 10:38.180 --> 10:40.060 network, has nothing to do with bits. 10:40.060 --> 10:46.940 But it is, like, Bitnet was a network also just for sending messages 10:46.940 --> 10:47.920 back and forth. 10:48.620 --> 10:52.300 And the first list server, actually, that was on that, like, 10:52.320 --> 10:57.600 distributing information to a lot of subscribers, the first of those 10:57.600 --> 10:59.240 list servers was Theorynet. 10:59.500 --> 11:02.680 That means the people in theoretical computer science were the first 11:02.680 --> 11:06.080 to actually make extensive use of the capabilities of such a network. 11:06.740 --> 11:15.140 And then in 81, the Computer Science Net was built, another backbone, 11:15.740 --> 11:23.900 separate from the ARPANET, to connect computer science departments at 11:23.900 --> 11:28.560 sites that had not so far an access to the ARPANET. 11:28.560 --> 11:34.620 And they immediately also planned for having interconnection between 11:34.620 --> 11:36.060 the CSNet and the ARPANET. 11:36.940 --> 11:38.960 This was all in the United States. 11:39.120 --> 11:41.340 In Europe, there was nothing like that. 11:41.640 --> 11:44.980 They had 213 sites at that time. 11:45.540 --> 11:52.440 In 83, the TCPIP became the standard protocol of ARPANET, 10 years 11:52.440 --> 11:53.940 after they started to design that. 11:53.940 --> 11:59.060 And at that year, in 83, they actually started the European Academic 11:59.060 --> 12:03.120 and Research Network, the ERN network, which was established not 12:03.120 --> 12:08.000 because the European academics were progressing very much and pushing 12:08.000 --> 12:12.600 for that, but IBM needed the technology that they had in the States. 12:12.960 --> 12:16.740 They wanted to have that in Europe to connect all their sites there. 12:17.040 --> 12:22.440 And so they provided it for free to the academic institutions in 12:22.440 --> 12:22.680 Europe. 12:22.680 --> 12:26.680 And so in Europe, we got the European Academic and Research Network. 12:27.360 --> 12:30.820 So this was something which we actually got at that time. 12:31.800 --> 12:39.040 And so in 83, I was actually as a guest or as a visiting researcher, a 12:39.040 --> 12:44.080 visiting professor in Canada, and there I followed discussions on how 12:44.080 --> 12:49.940 to actually establish or how to create first-year or how to supply 12:49.940 --> 12:52.540 first -year students with computers, what would be the best computer 12:52.540 --> 12:54.460 equipment with respect to these things. 12:56.060 --> 12:59.840 A discussion that was going on between all the computer science 12:59.840 --> 13:01.220 departments in North America. 13:01.860 --> 13:04.980 At the same time, in Europe, nobody would think about that. 13:05.060 --> 13:10.500 I came back in 84 to Germany and talked to the computing center head 13:10.500 --> 13:13.280 and asked for access to the networks. 13:13.360 --> 13:15.140 And they said, why do you need access to the network? 13:15.140 --> 13:19.260 I said, well, to send email, to follow news groups. 13:19.800 --> 13:21.020 And he said, well, you can phone. 13:21.400 --> 13:22.740 Why do you need such a network? 13:23.100 --> 13:24.460 That's not really necessary. 13:25.100 --> 13:27.360 And you don't use the computer for communication. 13:27.500 --> 13:29.680 You use it for computation and for nothing else. 13:30.320 --> 13:36.220 And so that was the attitude of typical computer science or computing 13:36.220 --> 13:38.400 center directors at that time. 13:38.400 --> 13:43.880 Although at Karlsruhe, at the same time, they received the first email 13:43.880 --> 13:46.660 message and supported that quite strongly. 13:46.860 --> 13:49.520 So there was a different attitude in different areas. 13:49.920 --> 13:54.400 So that's 30 years ago or almost. 13:55.340 --> 14:00.880 And so at that time, we actually got something which was similar in 14:00.880 --> 14:01.600 Europe. 14:02.480 --> 14:05.040 And then all these different things developed. 14:05.040 --> 14:08.380 You see how the number of hosts actually increased quite 14:08.380 --> 14:08.980 significantly. 14:09.020 --> 14:11.580 Here in 87, already 28,000 hosts. 14:12.280 --> 14:16.920 And they also had, at that time, stronger backbone lines of 1.5 14:16.920 --> 14:20.200 megabit per second, not just these 50 kilobit per seconds. 14:20.800 --> 14:27.220 And so all the technology improved, like the hardware, the 14:27.220 --> 14:32.860 communication technology, and also the software for running these 14:32.860 --> 14:33.280 things. 14:36.400 --> 14:38.880 Then in... oh, by the way, here, there was one thing. 14:38.960 --> 14:40.620 The Internet Engineering Task Force. 14:40.780 --> 14:43.960 I told you that there was a definition of the internet made up by the 14:43.960 --> 14:45.440 ITF. 14:45.720 --> 14:49.560 So they are actually controlling the ARPANET. 14:49.700 --> 14:52.000 They are the governing body of the internet. 14:52.860 --> 15:01.760 But it's all originally designed because of some requirements stated 15:01.760 --> 15:03.980 by the Department of Defense in the United States. 15:04.080 --> 15:06.320 So that was the major reason for setting up the ARPANET. 15:06.980 --> 15:10.040 So this developed on and on. 15:10.040 --> 15:18.420 So then the NSFnet, there were a few remarks on NSFnet. 15:18.720 --> 15:21.740 NSF is the National Science Foundation of the United States. 15:22.340 --> 15:27.040 It actually replaced the original ARPANET. 15:28.660 --> 15:34.020 And it was separated into a military network. 15:34.020 --> 15:38.840 So when you talk about hosts here, I would assume that this is always 15:38.840 --> 15:41.580 only the civil nodes and not the military nodes. 15:41.680 --> 15:45.740 Because the number of military nodes will not be public, probably. 15:47.620 --> 15:54.480 Then in the beginning of the 90s, we had actually here Tim Berners-Lee 15:54.480 --> 15:57.140 all of a sudden in Europe, in Geneva. 15:57.440 --> 16:01.900 And Tim Berners-Lee was a physicist, and he wanted to communicate with 16:01.900 --> 16:02.420 the colleagues. 16:02.420 --> 16:07.300 And so he designed a hypertext system for communication between 16:07.300 --> 16:10.060 researchers in high-energy physics. 16:10.600 --> 16:16.700 And so he designed hypertext, HTML, HTTP, and so on. 16:17.060 --> 16:19.620 That was in the beginning of the 90s. 16:20.160 --> 16:26.280 And in 92, actually, the World Wide Web was released as a service on 16:26.280 --> 16:26.900 the Internet. 16:26.920 --> 16:28.880 It was released by CERN. 16:28.880 --> 16:34.680 So it's an initiative of people in nuclear physics, not something 16:34.680 --> 16:40.320 coming from the military, not from computer scientists, but from 16:40.320 --> 16:40.720 physicists. 16:41.560 --> 16:44.780 And so this was the start of the World Wide Web. 16:45.080 --> 16:54.080 And then this really got momentum and evolved into what we now all 16:54.080 --> 16:55.840 know as the World Wide Web. 16:57.540 --> 17:06.280 Winton Cerf also, no, Tim Berners-Lee designed also the first browser 17:06.280 --> 17:08.000 for HTML pages. 17:09.040 --> 17:13.280 But then there was also the, like in the States, the Mosaic browser 17:13.280 --> 17:14.840 was developed. 17:15.660 --> 17:20.820 And so there was a standard interface for looking at content in the 17:20.820 --> 17:21.420 World Wide Web. 17:21.420 --> 17:26.600 And you see at that time there were already 2 million hosts, so quite 17:26.600 --> 17:27.580 a large number of hosts. 17:27.800 --> 17:33.900 And then this actually developed quite fast. 17:34.260 --> 17:40.320 So in 94, there was the first version of the Netscape browser. 17:41.560 --> 17:43.460 There were the first commercial applications. 17:44.740 --> 17:50.040 Pizza Hut was the first, or one of the first who used the World Wide 17:50.040 --> 17:51.960 Web for ordering pizzas. 17:52.480 --> 17:53.900 There was the first cyber bank. 17:54.760 --> 18:01.480 And so certain commercial applications actually developed here. 18:01.480 --> 18:08.900 And then when that was developing, countries got aware of, well, 18:08.960 --> 18:13.840 something is going on there, and we have to be, or we have to look at 18:13.840 --> 18:15.620 what's actually communicated there. 18:16.020 --> 18:19.740 So in China, certain restrictions were enforced. 18:19.920 --> 18:25.580 In Germany, certain, like access to certain use group was cut off. 18:25.580 --> 18:31.240 Certainly, like we have certain information which we consider to be 18:31.240 --> 18:36.420 unwanted on the Internet, like Nazi information and things like that, 18:36.600 --> 18:39.140 usually is something which is discarded. 18:40.120 --> 18:44.320 So it's not the case that we have just open information for everybody, 18:44.440 --> 18:48.000 but we also restrict it, but in a different way, I would say, than 18:48.000 --> 18:49.360 others do that. 18:49.360 --> 18:52.800 Let us see several things that are happening here in different 18:52.800 --> 18:55.320 countries with respect to the use of the Internet. 18:56.200 --> 19:04.220 And here in 98, you actually hear the Internet was used to judge the 19:04.220 --> 19:09.600 performance of world champion ice skaters as jurors there, so 19:09.600 --> 19:14.860 something which was impossible before that. 19:14.860 --> 19:19.840 The first electronic postal stamps were invented there by the U.S. 19:19.960 --> 19:20.540 Postal Service. 19:20.620 --> 19:22.100 We can do that now also in Germany. 19:23.920 --> 19:27.560 And, well, the backbones for the Internet actually had been 19:27.560 --> 19:32.620 transformed into private enterprises, so private companies actually 19:32.620 --> 19:36.500 are operating the backbones, which initially were operated by 19:36.500 --> 19:39.780 government institutions and not by private institutions. 19:42.140 --> 19:45.060 Then more things happened here. 19:45.600 --> 19:52.060 Viruses came up, like attacks on the network. 19:52.880 --> 19:56.160 The year 2000 was something which was very interesting because all 19:56.160 --> 20:00.940 people were concerned with respect to the time change or the date 20:00.940 --> 20:01.560 change. 20:02.080 --> 20:05.120 And it's really funny that in particular the U.S. 20:05.300 --> 20:10.140 timekeeper, like an official institution responsible for actually 20:10.140 --> 20:15.080 providing information on the correct time, they announced the new year 20:15.080 --> 20:20.960 as 19-100, the year 2000, because they did not actually manage to 20:20.960 --> 20:23.600 transform their system from just counting up. 20:24.340 --> 20:28.840 And after 99, the next one is 100, and they just added the first two 20:28.840 --> 20:32.220 digits, 19, so it was the year 19-100. 20:32.220 --> 20:37.520 And something which happened to a few time services around the world, 20:37.860 --> 20:45.420 but it didn't really come to all the problems that people were afraid 20:45.420 --> 20:48.040 of before this date change. 20:49.240 --> 20:52.360 But they had massive denial-of-service attacks. 20:52.360 --> 20:58.060 You know that here also the computing centers and universities have to 20:58.060 --> 21:04.400 do quite a bit to fight service attacks, like denial-of-service 21:04.400 --> 21:07.580 attacks, which is something which is happening again and again. 21:08.580 --> 21:13.740 In Europe, the European Commission actually decided to build up a very 21:13.740 --> 21:20.720 powerful backbone network, GEON, is the gigabit research network, so 21:20.720 --> 21:24.880 quite a change from the initial 50 kilobit per second to gigabit 21:24.880 --> 21:30.100 networks, but now we have all that. 21:31.400 --> 21:36.240 Then in 2001, certain other things happened, I don't want to go into 21:36.240 --> 21:41.520 all the individual things here, like city at home, something which is 21:41.520 --> 21:45.840 also interesting, using the computers that are connected to the 21:45.840 --> 21:51.300 network in order to distribute computations just in a very self 21:51.300 --> 21:52.020 -organized way. 21:52.140 --> 21:56.680 So city search for extraterrestrial intelligence is just sending 21:56.680 --> 22:06.800 information to computers, and then when the screensaver is running, 22:08.060 --> 22:21.000 the screensaver actually is performing analysis on certain parts of 22:21.000 --> 22:26.160 the communication, of the information that is recorded by certain 22:26.160 --> 22:29.920 recorders who look for signals from outer space. 22:31.260 --> 22:32.780 Okay, what else? 22:34.480 --> 22:43.660 Here, there was all kinds of attacks on the internet. 22:43.660 --> 22:49.540 This is here just a global network for advanced application 22:49.540 --> 22:50.960 development. 22:51.500 --> 22:56.480 This is really global between Chicago, Amsterdam, Moscow, like Russian 22:56.480 --> 22:58.540 states, Russian and China. 23:01.080 --> 23:05.500 In 2004, for the first time, there are more instances of domain name 23:05.500 --> 23:10.480 servers outside the US, so this is something which, certainly as we 23:10.480 --> 23:14.080 know, is now very widespread all over the world. 23:15.220 --> 23:17.520 And what else? 23:17.640 --> 23:23.180 New domains have been initiated there. 23:25.900 --> 23:31.800 Certain things happened, I don't want to go into too many things 23:31.800 --> 23:32.080 there. 23:32.080 --> 23:33.640 Here, this is also interesting. 23:33.840 --> 23:36.900 Twitter is asked by the US government to delay plant maintenance of 23:36.900 --> 23:37.460 its service. 23:38.520 --> 23:41.880 As a result of heavy use by Iranian users, you know that you had the 23:41.880 --> 23:47.780 Arabian Spring events, which were heavily influenced by the 23:47.780 --> 23:49.560 communication capabilities of the internet. 23:50.020 --> 23:54.460 So you know all that, I guess, and I just wanted to briefly show you a 23:54.460 --> 23:57.080 few diagrams on how the internet actually grew. 23:57.080 --> 24:01.580 So the number of hosts, starting with four hosts, as I mentioned, and 24:01.580 --> 24:06.460 then we have almost a doubling every year, and now we are somewhere 24:06.460 --> 24:11.340 around, like almost at 900 million, so the last figure here was from 24:11.340 --> 24:12.920 September 11. 24:13.460 --> 24:17.740 This is an update of the slides that I have put on the server before, 24:18.300 --> 24:20.560 so just a few years more in there. 24:20.560 --> 24:22.460 It's just growing and growing and growing. 24:23.680 --> 24:28.920 So it's really a very, very large number of hosts that are connected 24:28.920 --> 24:29.500 to the internet. 24:30.540 --> 24:34.520 And then if you look at the number of domains, that's also growing. 24:34.520 --> 24:40.020 The number of websites, there you see strange jumps. 24:40.300 --> 24:45.560 These jumps have to do with certain providers, like service providers, 24:47.660 --> 24:54.620 offering websites going out of service or into service, and so here 24:54.620 --> 24:59.900 sometimes things are counted or recounted, and then they notice that 24:59.900 --> 25:06.900 there are certain changes in the number of sites that are actually 25:06.900 --> 25:07.240 there. 25:07.460 --> 25:12.860 The most important thing is that certainly there have been some minor 25:12.860 --> 25:16.640 changes there, for example the number of websites, you don't see 25:16.640 --> 25:25.160 actually the downturn or the drop in internet activities after the 25:25.160 --> 25:32.600 year, like the early years of the 2000s, we had like the internet 25:32.600 --> 25:38.480 bubble exploded and the activities went down, but you don't see it 25:38.480 --> 25:43.980 here actually, and then we had the economic problems after the year 25:43.980 --> 25:44.740 2008. 25:45.180 --> 25:47.540 You see a little bit here, but not that much. 25:47.860 --> 25:51.080 So with respect to the internet, with respect to communication, it's 25:51.080 --> 25:54.800 just an increasing thing. 25:55.520 --> 26:00.100 So regardless what you look at, even here if you look at the number of 26:00.100 --> 26:05.580 Facebook accounts, the slide which I inserted last night, you also see 26:05.580 --> 26:07.880 that it's just continually growing. 26:08.080 --> 26:13.140 So Facebook accounts are almost 900 million Facebook accounts. 26:13.340 --> 26:17.840 I think I recently read something that they have reached 1 billion 26:17.840 --> 26:20.080 Facebook accounts meanwhile. 26:20.080 --> 26:25.700 So it's an amazing number of people connected to internet and 26:25.700 --> 26:35.060 definitely this has quite some influence on how society and industry 26:35.060 --> 26:35.760 is developing. 26:36.340 --> 26:37.540 What are the reasons for that? 26:37.540 --> 26:41.700 Certainly one reason is that we have this tremendous growth of 26:41.700 --> 26:43.280 bandwidth and wide area networks. 26:44.120 --> 26:50.200 So we are now in the gigabit range and people are actually working on 26:50.200 --> 26:51.400 terabit networks. 26:52.580 --> 26:56.420 So we can communicate or we can transport all the information that is 26:56.420 --> 26:59.900 necessary over those wide area networks. 26:59.900 --> 27:06.400 And then we also have, like in the local area networks, on the 27:06.400 --> 27:10.040 telephone exchange systems, we have also improved transmission 27:10.040 --> 27:10.600 performance. 27:10.820 --> 27:14.280 Initially people thought that, well, you can just communicate with 27:14.280 --> 27:18.100 several thousand bits per second, but not really more. 27:18.420 --> 27:20.760 We know that we have quite some development there. 27:20.760 --> 27:25.160 So initially we had here, like the first step forward was the ISDN 27:25.160 --> 27:29.720 network providing 64 kilobits per second in Germany with several 27:29.720 --> 27:30.960 services on top of that. 27:31.300 --> 27:37.460 It was an important reason for actually allowing more communication 27:37.460 --> 27:38.640 over telephone lines. 27:38.640 --> 27:43.500 And the major push was the asymmetric digital subscriber line 27:43.500 --> 27:48.640 technology, which is just some breakthrough because it can be used on 27:48.640 --> 27:54.000 the same lines that before had been considered to have a natural limit 27:54.000 --> 27:55.880 of a few thousand bits per second. 27:56.420 --> 28:02.280 And now with the ADSL technology, you could actually send millions of 28:02.280 --> 28:04.840 bits per second as we know in DSL connections. 28:04.840 --> 28:10.080 We are at about 10, 20, 30 megabit per second. 28:10.420 --> 28:17.780 And this is just due to a different way of coding information over 28:17.780 --> 28:20.740 those lines. 28:20.820 --> 28:24.860 So it's a different way of coding information and this just allows to 28:24.860 --> 28:29.580 increase the information that is broadcasted or that is sent over a 28:29.580 --> 28:30.620 line so significantly. 28:31.220 --> 28:36.200 One thing is that it is asymmetric, so we have like the downstream 28:36.200 --> 28:41.400 connection is significantly larger than the upstream connection. 28:41.600 --> 28:46.640 Upstream is just a few megabit per second while downstream can be in 28:46.640 --> 28:48.240 the range of 100 megabit per second. 28:48.760 --> 28:51.460 So this is quite a difference, but normally you are just receiving 28:51.460 --> 28:53.940 information, you're not sending that much information. 28:54.280 --> 28:57.480 So it's perfect for a network which is mainly distributing 28:57.480 --> 28:57.980 information. 28:58.960 --> 29:03.580 Okay, and now we have meanwhile also UMTS and LTE. 29:03.740 --> 29:07.840 LTE for long-term evolution, the new standard that is just being 29:07.840 --> 29:10.620 established in most countries of the world. 29:10.800 --> 29:14.560 It's not available everywhere, so long-term evolution is a standard 29:14.560 --> 29:17.680 that's called the 4G, fourth generation standard for mobile 29:17.680 --> 29:18.260 communication. 29:18.260 --> 29:26.560 And the bandwidth there is around 300 megabits per second, like the 29:26.560 --> 29:31.020 downlink peak rate and the uplink peak rate around 75 megabit per 29:31.020 --> 29:31.280 second. 29:31.580 --> 29:36.640 So this is at the same range as you have for wired communication. 29:36.640 --> 29:41.920 So it means that with mobile communication, with mobile devices, we 29:41.920 --> 29:45.040 have the same communication bandwidth as we do with wired 29:45.040 --> 29:45.660 communication. 29:46.040 --> 29:51.040 And this is certainly very important for many services that are used 29:51.040 --> 29:54.220 in communication and computation. 29:54.840 --> 29:59.000 So all the cloud services and so on can be used much better if you 29:59.000 --> 30:00.320 have such a technology available. 30:00.320 --> 30:04.840 So this is really a major step, although the name is a bit strange, 30:05.160 --> 30:06.560 LTE, long-term evolution. 30:06.700 --> 30:11.280 Nobody would think about a connection to a communication standard, but 30:11.280 --> 30:13.000 that's how things just developed. 30:13.360 --> 30:19.380 Some people came up with that name, and it stands for this new, very 30:19.380 --> 30:24.320 powerful technology for sending information in mobile communication. 30:25.110 --> 30:28.200 And then, like this was the increased performance of our telephone 30:28.200 --> 30:28.660 lines. 30:28.940 --> 30:33.300 And then the other important part certainly is that we have 30:33.300 --> 30:36.720 standardized platform-independent interfaces to the World Wide Web. 30:37.160 --> 30:39.600 More or less platform-independent, we know that there are certain 30:39.600 --> 30:41.100 differences between different browsers. 30:42.780 --> 30:44.500 So this is annoying. 30:44.500 --> 30:49.560 The original idea was to have just one standardized way of displaying 30:49.560 --> 30:56.020 information, but companies have their own view on that and wanted to 30:56.020 --> 31:00.040 have some competitive advantage, so do things different from others. 31:00.660 --> 31:04.660 And so you have now, if you provide information, you have to check how 31:04.660 --> 31:07.000 it is actually displayed on the different platforms. 31:07.000 --> 31:10.600 You also have the programming language Java, which supports just 31:10.600 --> 31:15.800 having all this way of providing information. 31:16.320 --> 31:21.300 But you should always remember that there's the history of the 31:21.300 --> 31:27.440 Internet, which is about 40 years, and there's also the history of the 31:27.440 --> 31:29.180 World Wide Web, which is just 20 years. 31:29.840 --> 31:32.400 So the World Wide Web is just a service on the Internet. 31:32.400 --> 31:37.040 So if we talk about Internet, very often we mean the World Wide Web, 31:37.640 --> 31:40.860 but we're actually talking about the World Wide Web, and so the 31:40.860 --> 31:43.740 Internet is something which is independent of the World Wide Web. 31:45.700 --> 31:49.780 Just a remark on what I showed you here very briefly, so you can read 31:49.780 --> 31:53.240 it if you have time, just in more detail. 31:54.520 --> 31:58.840 What I present to you here is information I got from the Internet, 31:59.040 --> 32:02.300 from certain websites on history of the Internet, and it's mainly 32:02.300 --> 32:06.180 showing the American view on the Internet, although it is something 32:06.180 --> 32:09.740 which has been influenced heavily by the North Americans. 32:10.480 --> 32:14.920 And so Europe has been more involved in the World Wide Web, and so the 32:14.920 --> 32:19.740 World Wide Web Consortium, which is ruling, governing the World Wide 32:19.740 --> 32:22.280 Web, is something which is located in Europe. 32:24.500 --> 32:27.720 And it's interesting that the Internet and the World Wide Web evolved 32:27.720 --> 32:31.620 in universities and research institutes, and it took very long before 32:31.620 --> 32:36.080 the industry actually recognized the advantages and the economic 32:36.080 --> 32:37.180 potential of the Internet. 32:37.380 --> 32:41.600 So this is something which was not, as sometimes people say, industry 32:41.600 --> 32:45.920 is just showing where technology needs to, this is something which was 32:45.920 --> 32:51.580 designed actually in academia, and as I showed you, theoretical 32:51.580 --> 32:56.880 physicists or nuclear physicists were very important there, and also 32:56.880 --> 32:59.740 the theoretical computer scientists, they made use of those 32:59.740 --> 33:03.940 technologies, and it's not always those who are, who should be 33:03.940 --> 33:07.240 considered to be those who are in front of that development. 33:07.780 --> 33:12.780 So here it also says that now we have, that we certainly have gigabit 33:12.780 --> 33:16.260 networks, there are experiments with terabit networks, there are 33:16.260 --> 33:19.420 strong groups in that area here at Karlsruhe, the KRT, like the groups 33:19.420 --> 33:27.700 of Jörg Leuthold and Professor Coase, they just, they have won several 33:27.700 --> 33:34.400 prizes for that, because they have designed very, very powerful high 33:34.400 --> 33:37.200 -speed and very efficient communication networks. 33:37.820 --> 33:41.560 So we are in the range of terabit networks in research, and certainly 33:41.560 --> 33:47.960 it means that our computers have to be able to actually process the 33:47.960 --> 33:52.220 information that you can move over those backbone lines. 33:52.460 --> 33:57.060 You have to be able to process the information at the same speed, and 33:57.060 --> 34:00.120 so you need very powerful computers for that. 34:00.800 --> 34:04.720 And you know that one important topic meanwhile is, since we have that 34:04.720 --> 34:10.540 much information, we are talking in several areas about a topic which 34:10.540 --> 34:12.160 is abbreviated as big data. 34:12.160 --> 34:17.380 And big data is coming from certainly experiments where you just 34:17.380 --> 34:22.900 produce lots of data, terabits or petabits of data a day. 34:23.260 --> 34:29.020 For example, in the Large Hadron Collider experiments at Geneva, or 34:29.020 --> 34:33.080 here at Karlsruhe in the KRT, there are certain experiments where they 34:33.080 --> 34:39.240 analyze the development of, or the evolution of zebrafishes, small 34:39.240 --> 34:44.760 fishes, which show interesting developments, interesting evolutionary 34:44.760 --> 34:45.180 steps. 34:45.820 --> 34:53.240 And this is very similar to what's happening in Stamtzell. 34:53.600 --> 34:57.460 I don't know the right word for Stamtzell in the moment, so it's the 34:57.460 --> 35:00.960 basic, the core cells, the original stem cells. 35:01.260 --> 35:04.100 It's just the stem cells. 35:04.100 --> 35:07.460 And a lot of Stamtzell research is done here with zebrafishes, and 35:07.460 --> 35:08.780 it's millions of zebrafishes. 35:08.940 --> 35:14.260 They also have lots of information that has to be processed and has to 35:14.260 --> 35:15.980 be made available for many different people. 35:16.340 --> 35:18.740 And this can be made available using those networks. 35:19.280 --> 35:26.720 Okay, that's some remarks on the general Internet history, and now a 35:26.720 --> 35:28.420 few remarks on the German Internet history. 35:28.420 --> 35:33.680 As I told you that we had, in 83, we got the European Academic 35:33.680 --> 35:34.480 Research Network. 35:35.400 --> 35:39.480 And so in the mid-80s, in Germany, we got the Deutsches 35:39.480 --> 35:41.800 Forschungsnetz, the German Research Network. 35:42.400 --> 35:47.420 And it was established to provide remote access to computers. 35:47.680 --> 35:51.560 So it was only there for getting remote access to perform computations 35:52.640 --> 35:55.560 at distant computers. 35:55.560 --> 35:57.140 This was not for communication. 35:57.660 --> 36:00.780 It was not supporting e-mail services and things like that. 36:00.820 --> 36:04.900 It was only there for doing remote log-in and performing computations 36:04.900 --> 36:08.340 at different computing centers. 36:08.620 --> 36:09.740 Not for communication. 36:10.340 --> 36:15.500 It was based on a protocol called X-25, just a telephone line 36:15.500 --> 36:15.820 protocol. 36:16.740 --> 36:21.180 And then later on, in the end of the 90s, this was transformed into 36:21.180 --> 36:25.500 the Science Network, the Wissenschaftsnetz, and actually upgraded to 36:25.500 --> 36:27.580 35 megabits per second in the backbones. 36:27.840 --> 36:34.060 And then a few years later, we got the Gigabit Science Network. 36:34.860 --> 36:41.600 And now we have a widely distributed network having around 10 gigabit 36:41.600 --> 36:41.920 lines. 36:42.020 --> 36:43.060 So now it's very powerful. 36:43.220 --> 36:44.860 It's as powerful as in other countries. 36:45.400 --> 36:50.640 So we have been behind the North American developments for many years. 36:50.640 --> 36:55.900 But this changed at the end of the 90s because people got more aware 36:55.900 --> 36:58.400 of the necessity of having that. 36:58.520 --> 37:03.240 At those times, our research policy, our research funding was on 37:03.240 --> 37:09.560 nuclear science, on large-scale research, not on small-scale research, 37:09.940 --> 37:13.820 or small-scale with respect to microelectronics and things like that, 37:13.960 --> 37:15.260 and not on communication infrastructure. 37:15.260 --> 37:21.340 It was just a wrong focus on... 37:21.340 --> 37:26.020 or a wrong strategic decision on what kind of research is necessary in 37:26.020 --> 37:26.540 this country. 37:27.500 --> 37:34.120 And a very interesting thing was that in the 90s, in the mid-90s, in 37:34.120 --> 37:38.960 the States, there was Al Gore, Vice President, who had as his topic 37:38.960 --> 37:42.700 the Information Superhighway Research Program, a very hot topic in the 37:42.700 --> 37:43.220 United States. 37:43.220 --> 37:48.240 And when that was at the same time discussed in the German government, 37:49.800 --> 37:53.860 the Chancellor at that time, the mid-90s, said, okay, this should be 37:53.860 --> 37:55.460 sent to the traffic... 37:55.460 --> 37:58.580 or to the committee for traffic, because it's on highways. 37:59.540 --> 38:03.000 And that was the way this topic was dealt with. 38:03.200 --> 38:06.960 So they did not see, actually, the importance of that infrastructure 38:06.960 --> 38:10.600 for the development of society, and in particular, for industry. 38:11.400 --> 38:13.360 And so this is... 38:13.360 --> 38:20.020 this changed in the end of the 90s, but in the mid of the 90s, when 38:20.020 --> 38:24.880 the World Wide Web came up, the politicians were afraid of threat of 38:24.880 --> 38:26.160 sex and crime on the Internet. 38:26.480 --> 38:31.540 They made suggestions like, well, it might be okay if content like 38:31.540 --> 38:37.500 this content is available after 11 p.m. 38:37.540 --> 38:38.140 at night. 38:38.140 --> 38:44.080 It may be okay, but not before that, because then it may be dangerous 38:44.080 --> 38:45.800 for kids who look at that. 38:46.320 --> 38:49.160 They did not actually see that on the Internet you don't talk about 38:49.160 --> 38:49.480 time. 38:49.900 --> 38:51.900 It's just there, or not there. 38:52.480 --> 38:56.200 And it is independent of at what time in one country it is available 38:56.200 --> 38:56.700 or visible. 38:57.380 --> 38:58.980 So this was strange. 38:59.260 --> 39:02.840 This was something which was just the same all over the different 39:02.840 --> 39:03.660 political parties. 39:04.540 --> 39:08.700 And this changed in the end of the 90s, and nowadays we know we are 39:08.700 --> 39:11.740 aware of what's necessary there. 39:12.260 --> 39:13.400 At least to some extent. 39:13.620 --> 39:15.340 Just one more remark on that. 39:15.860 --> 39:20.420 There is a traditional area of research, like energy-related research 39:20.420 --> 39:21.060 in Germany. 39:23.060 --> 39:25.060 It has a long tradition. 39:25.700 --> 39:28.180 People talk about energy efficiency and things like that. 39:29.000 --> 39:35.800 And we all know that due to the changes in the way we generate power, 39:35.800 --> 39:41.140 electrical power, from renewable sources, we need to change the 39:41.140 --> 39:42.020 management systems. 39:42.840 --> 39:47.020 But that means that we need information and communication technology 39:47.020 --> 39:48.220 in energy research. 39:48.740 --> 39:52.660 And this is not really seen by those in the government who are 39:52.660 --> 39:55.020 responsible for energy research. 39:55.260 --> 39:59.120 They deny the need for research on information and communication 39:59.120 --> 39:59.680 technologies. 39:59.980 --> 40:04.000 It's other areas, other parts of the government who are related to 40:04.000 --> 40:07.680 information and communication technologies which say we have to do 40:07.680 --> 40:08.680 something in that area. 40:08.840 --> 40:13.320 So they design their own programs on ICT for energy, but those who are 40:13.320 --> 40:17.980 responsible for energy research so far deny that necessity. 40:18.780 --> 40:21.760 So this is something which has some tradition in Germany, that 40:21.760 --> 40:27.220 information and communication technology is not really seen as what it 40:27.220 --> 40:32.640 really is in the moment, mainly the major driver and enabler of new 40:32.640 --> 40:35.880 technologies, of innovation. 40:36.680 --> 40:40.260 And so I see I should speed up a bit. 40:41.380 --> 40:46.040 But I think it's also necessary to tell you these things that are 40:46.040 --> 40:52.280 going on apart from the research topics that we have here. 40:53.280 --> 40:56.800 So, now we come to more content. 40:56.800 --> 41:02.040 I said that the Internet is made up of addresses, protocols, and 41:02.040 --> 41:02.500 services. 41:02.960 --> 41:04.840 What are the standard services that we have? 41:05.420 --> 41:09.480 For example, the first one is a virtual terminal for remote log-in. 41:10.340 --> 41:11.640 That was the initial thing. 41:11.700 --> 41:15.180 You would like to log into another computer over a network. 41:15.640 --> 41:19.720 Then email, also one of the first services to actually provide 41:19.720 --> 41:22.540 communication, to send messages in a very simple way. 41:22.540 --> 41:26.580 The simple mail transfer protocol was designed in the early days of 41:26.580 --> 41:27.000 the Internet. 41:27.600 --> 41:30.060 File transfer protocol, also an important thing. 41:30.480 --> 41:33.620 News groups were also among the initial services. 41:33.920 --> 41:37.220 And then the World Wide Web, as you know, made up of several different 41:37.220 --> 41:42.500 architectures, or several different protocols and languages, the HTML, 41:43.960 --> 41:48.920 markup language, the hypertext transfer protocol, and the way we 41:48.920 --> 41:54.300 actually address the different resources in the World Wide Web, the 41:54.300 --> 41:57.680 URL concept was designed at those times. 41:58.000 --> 42:01.460 And all these together actually created the World Wide Web. 42:02.040 --> 42:06.720 And we have all kinds of Internet technologies, like company-wide 42:06.720 --> 42:08.840 networks based on the Internet technology. 42:09.340 --> 42:13.620 For a long time there were very proprietary networks, network 42:13.620 --> 42:15.200 protocols in different companies. 42:15.440 --> 42:19.860 And so you had internal protocols, you had external protocols, and the 42:19.860 --> 42:23.780 fact that they are now built on the same technology certainly 42:23.780 --> 42:27.160 facilitates the way we can actually manage those systems. 42:28.140 --> 42:30.480 So, now we come to Internet technology. 42:30.780 --> 42:36.440 First some small view on what actually is the structure. 42:36.620 --> 42:40.740 So the backbone here is the high bandwidth Internet kernel, several 42:40.740 --> 42:44.820 strong lines connecting the major sites and transporting information 42:44.820 --> 42:50.760 between different larger sites in the network, having high capacity 42:50.760 --> 42:51.900 and high speed cables. 42:52.520 --> 42:57.120 Although certainly, like the latency to send information over a long 42:57.120 --> 43:01.380 distance is still there, but the number of bits you can send in 43:01.380 --> 43:05.120 parallel has increased, so that way we can also increase the bandwidth 43:05.120 --> 43:07.720 of those communication lines. 43:08.020 --> 43:11.180 But you always have to be aware of the fact that it takes some time to 43:11.180 --> 43:14.520 actually send information between different locations. 43:15.180 --> 43:19.360 You know that there is one technology which is very promising, or at 43:19.360 --> 43:23.240 least that is very interesting. 43:23.600 --> 43:27.680 It is the quantum communication or quantum computing. 43:28.680 --> 43:33.380 But quantum communication is very promising because there you would 43:33.380 --> 43:40.000 have communication in zero time, because if you change something here 43:40.000 --> 43:42.840 at the same time it is changed there, and so you can have 43:42.840 --> 43:45.440 instantaneous communication between different sites. 43:45.780 --> 43:50.100 This is something which researchers mainly in physics are working on, 43:50.660 --> 43:57.140 and progress in that area certainly will be very important for future 43:57.140 --> 43:58.320 communication systems. 43:58.760 --> 44:02.220 Okay, then we have the wide area networks, we have metropolitan area 44:02.220 --> 44:07.240 networks, for example, in larger municipal areas, larger urban areas, 44:07.820 --> 44:16.940 and we have the local area networks, and then certainly all these 44:16.940 --> 44:21.520 networks have to be connected, so we have network access points for 44:21.520 --> 44:28.460 the different computers that are connected in some way to different 44:28.460 --> 44:29.080 technologies. 44:29.300 --> 44:32.820 Here you have the Ethernet technology, there I indicated the token 44:32.820 --> 44:36.020 ring or the token bus technologies. 44:36.580 --> 44:40.000 Nowadays most people or most LANs are based on the Ethernet. 44:40.700 --> 44:44.200 I will not talk about token ring and Ethernet in this course. 44:44.660 --> 44:50.780 I will talk more about the technology that is above that, so above the 44:50.780 --> 44:51.680 network access points. 44:52.180 --> 44:56.080 Then we have routers between the different locations in the network, 44:56.560 --> 45:03.320 and we have gateways between different regions so that you can address 45:03.320 --> 45:08.740 a certain region and then refine that and go to certain locations in 45:08.740 --> 45:08.980 that. 45:09.140 --> 45:11.920 So if you have a different view of the Internet, we have certain 45:11.920 --> 45:16.660 autonomous systems which are connected in some way using gateways. 45:17.000 --> 45:21.160 We have the local area networks and we have routers in between, and so 45:21.700 --> 45:26.740 when we communicate, we can communicate either inside those networks, 45:26.860 --> 45:32.540 those autonomous systems, or go to some using some communication 45:34.240 --> 45:37.640 facility, go to different autonomous systems. 45:37.980 --> 45:43.980 Meanwhile we have hundreds of thousands of autonomous systems which 45:43.980 --> 45:48.360 are making up the complete Internet. 45:49.600 --> 45:54.580 One important point also is that the Internet is a network which is 45:54.580 --> 45:56.360 more or less self-organizing. 45:56.840 --> 46:02.460 There are certain specific protocols which have been designed and 46:02.460 --> 46:07.140 approved by the governing authorities, but a major point of that is 46:07.140 --> 46:11.900 that all the nodes in the network actually can carry out those 46:12.510 --> 46:16.760 protocols and can in some way control the network flow. 46:16.900 --> 46:21.300 This is nothing which is centrally organized, but it is distributed 46:22.040 --> 46:26.520 and so we have here essentially the Internet is a very large self 46:26.520 --> 46:27.340 -organized system. 46:27.960 --> 46:35.200 And how it is actually operating is we will look at that in a moment, 46:35.300 --> 46:40.780 looking at the different protocols in the communication that we have 46:40.780 --> 46:41.260 in the Internet. 46:41.640 --> 46:42.940 So we have different layers. 46:43.160 --> 46:49.260 There is the so-called open system interconnection network layer model 46:49.260 --> 46:51.160 of the International Standard Organization. 46:51.360 --> 46:53.080 It's a seven layer network. 46:53.540 --> 46:56.620 In the Internet we are only looking at five different layers 46:56.620 --> 46:57.280 essentially. 46:59.080 --> 47:01.020 And here I show only four of them. 47:01.120 --> 47:03.620 So we have the application layer where all the services are running. 47:04.080 --> 47:05.340 We have TCPIP. 47:05.720 --> 47:07.340 That's the protocol suite that I mentioned. 47:08.120 --> 47:11.420 There are some more protocols in there which usually are not mentioned 47:11.420 --> 47:11.940 explicitly. 47:11.940 --> 47:14.520 But later on, here's a question. 47:15.100 --> 47:15.260 Oops. 47:17.800 --> 47:21.900 Are the internal gateways the same as the routers on the previous 47:21.900 --> 47:22.360 slide? 47:22.920 --> 47:30.080 Like the routers and gateways are sometimes used almost in a 47:30.080 --> 47:30.940 synonymous way. 47:34.020 --> 47:41.960 There are some differences between gateways that are addressed if you 47:43.440 --> 47:46.980 go from one autonomous network to the next autonomous network. 47:47.220 --> 47:50.060 So there you have certain gateways. 47:50.180 --> 47:54.060 So gateways are special routers running certain special other 47:54.060 --> 47:57.080 protocols which we will briefly look at in a moment. 47:57.220 --> 48:01.840 So they are not exactly the same but they are very similar just having 48:01.840 --> 48:05.360 a few more tasks to do. 48:06.380 --> 48:06.580 Okay. 48:07.060 --> 48:08.320 Another question. 48:09.200 --> 48:11.260 That's the same question again. 48:12.300 --> 48:12.560 Okay. 48:13.440 --> 48:19.720 So we have these different layers and the question then is how that 48:19.720 --> 48:20.680 actually works. 48:20.960 --> 48:25.980 So we have here these protocols on different layers. 48:26.180 --> 48:32.840 So it's a separation of concerns of the different protocols and so we 48:32.840 --> 48:38.240 have a certain application for example Telnet or FTP or SMTP, HTTP, 48:38.720 --> 48:42.460 whatever you like an application which would like to send certain 48:42.460 --> 48:42.840 data. 48:43.320 --> 48:48.540 Now if you have, if you would like to send certain data then it may be 48:48.540 --> 48:50.280 a long stream of data. 48:50.900 --> 48:53.780 Now you cannot send just a stream of data. 48:53.920 --> 48:55.900 You have to organize the sending of data. 48:56.040 --> 48:59.820 I mentioned that the major part of the internet is that it is a packet 48:59.820 --> 49:01.280 switched network. 49:01.900 --> 49:06.760 So data has to be put into packets and the question is what kind of 49:06.760 --> 49:08.460 packet size do we actually need? 49:09.540 --> 49:16.500 Packet size will be determined by the capabilities of the actual nodes 49:16.500 --> 49:21.540 that are in the network and so you have to communicate information on 49:21.540 --> 49:23.720 what an appropriate packet size actually would be. 49:24.780 --> 49:28.320 And the first thing that is done is that the transmission control 49:28.320 --> 49:31.780 protocol is actually looking at that data. 49:32.180 --> 49:39.540 It is separating the data stream into segments and it is building up 49:39.540 --> 49:44.960 communication to the receiver between the sender and the receiver. 49:45.600 --> 49:47.520 So we have one communication line. 49:48.160 --> 49:52.740 Now the network, like here, if that is the sender and that is the 49:52.740 --> 49:55.680 receiver, there will be many nodes in between here. 49:56.160 --> 50:01.380 And having many different connections, so there might be many 50:01.380 --> 50:07.420 different possibilities to actually send information from one node to 50:07.420 --> 50:07.820 the other. 50:08.740 --> 50:12.020 The question is if we have here one line, what does that mean? 50:12.120 --> 50:13.640 This is just a logical connection. 50:13.860 --> 50:17.660 This is a communication that is sent between the sender and receiver. 50:18.120 --> 50:21.500 A few packets send back and forth and then they decide on being 50:21.500 --> 50:22.020 connected. 50:22.460 --> 50:26.260 This is just a logical connection and they know how they will actually 50:26.260 --> 50:27.720 treat the follow-up packets. 50:28.420 --> 50:30.980 So this is a logical connection, not a physical connection. 50:31.460 --> 50:34.720 That means that different packets may run over different lines. 50:35.720 --> 50:42.680 And then we have to provide certain information for the receiving 50:42.680 --> 50:43.140 side. 50:43.260 --> 50:50.380 That means when data is sent down here, like over this link on the 50:50.380 --> 50:54.400 other side, this information has to be looked at again. 50:55.260 --> 51:01.320 So every protocol, TCP, IP and then some protocol on the physical data 51:01.320 --> 51:05.680 link layer, they have to provide some information which is necessary 51:05.680 --> 51:10.280 for those on the other side actually to know what to do with this data 51:10.280 --> 51:10.600 item. 51:11.480 --> 51:18.420 And so the next hop there will look at the last header, gets 51:18.420 --> 51:20.420 information on what has to be done with that. 51:21.280 --> 51:26.760 And then it may provide the information to the next layer of that 51:26.760 --> 51:28.960 protocol stack, maybe to IP. 51:29.480 --> 51:33.800 IP looks at the header that has been moved, that has been appended 51:33.800 --> 51:35.480 there by the sending side. 51:36.120 --> 51:41.340 And then you get information on what is actually the destination of 51:41.340 --> 51:44.460 that packet, how should I treat that packet, certain information. 51:44.660 --> 51:46.060 We will look at that in a moment. 51:47.140 --> 51:52.300 And if it's actually the receiving side and not just some intermediate 51:52.300 --> 51:57.020 side where you always go up to the IP layer, if it's the receiving 51:57.020 --> 52:02.500 side, then IP will forward this datagram to the next layer to the TCP 52:02.500 --> 52:03.040 protocol. 52:03.920 --> 52:08.880 And then TCP here will look at the header that has been provided there 52:08.880 --> 52:13.560 by the sender and knows what to do with the data, how to assemble the 52:13.560 --> 52:17.560 data into the data for the application. 52:17.940 --> 52:23.400 So the data items here are merged into the data stream for the 52:23.400 --> 52:24.800 application on the receiving side. 52:26.180 --> 52:29.620 And the important points are that this is the separation of concerns. 52:29.860 --> 52:35.420 Different aspects are treated by or are considered by the different 52:35.420 --> 52:41.100 protocols and so you can have a systematic way of actually providing 52:41.100 --> 52:43.500 the communication between different sites. 52:44.260 --> 52:48.860 Okay, we will look at the different protocols in particular at IP and 52:48.860 --> 52:54.560 TCP later on, certainly also on the applications in the next lectures. 52:55.000 --> 53:00.460 So let's start with IP, the Internet protocol which is located on the 53:00.460 --> 53:05.040 network layer and so what does IP actually do? 53:05.720 --> 53:08.420 It defines first of all the addressing format in the Internet. 53:08.960 --> 53:13.800 So we know that we have a global addressing scheme in the Internet. 53:15.100 --> 53:22.060 And the task of the IP protocol is to actually let a node decide what 53:22.060 --> 53:27.060 to do with messages that are received from different locations and 53:27.060 --> 53:30.920 which would like to go to different destinations. 53:31.640 --> 53:36.620 And the question is what is the most appropriate link or the most 53:36.620 --> 53:41.160 appropriate next node where such information has to be sent to. 53:41.460 --> 53:44.760 So this is the major task that has to be looked at. 53:45.800 --> 53:51.180 The routers like this is a router has to know what to do with the 53:51.180 --> 53:55.360 datagram and so it gets some information and has to decide what is the 53:55.360 --> 53:59.400 most appropriate next hop on the path from the source to the 53:59.400 --> 53:59.940 destination. 54:01.160 --> 54:06.180 And it does that in an unreliable and connectionless way. 54:06.600 --> 54:08.420 Now why do we say unreliable? 54:08.500 --> 54:10.000 We always need reliability. 54:11.060 --> 54:16.140 What I state here is just that IP is not concerned with the 54:16.140 --> 54:16.560 reliability. 54:17.340 --> 54:21.840 It just decides on what to do with the packet which it receives, how 54:21.840 --> 54:24.020 to forward it to the next hop. 54:24.580 --> 54:28.140 It's not concerned with the question whether that datagram will 54:28.140 --> 54:33.360 actually be received sometime later by the destination host. 54:34.360 --> 54:40.680 And so in this way it is unreliable because it doesn't care about the 54:40.680 --> 54:46.340 fact whether the datagram is received by the supposed destination. 54:47.120 --> 54:48.560 And it's also connectionless. 54:48.560 --> 54:53.760 That means if you get a sequence of datagrams from maybe the same 54:53.760 --> 54:59.180 site, the same source directed to the same destination it may be that 54:59.180 --> 55:01.660 the messages are sent over different links. 55:02.060 --> 55:08.080 If certain things have changed in between certain traffic jams on some 55:08.080 --> 55:13.220 link, that means a delay in communication may lead to choosing a 55:13.220 --> 55:18.880 different connection for this sequence of messages, sequence of 55:18.880 --> 55:19.480 datagrams. 55:19.800 --> 55:24.040 So it's connectionless just a logical connection, not a physical 55:24.040 --> 55:24.580 connection. 55:25.900 --> 55:30.480 And the next layer above, the transport layer with the TCP protocol is 55:30.480 --> 55:35.180 then responsible for reliability but IP is just responsible for 55:35.180 --> 55:37.680 deciding on the next hop. 55:38.020 --> 55:38.500 That's it. 55:38.820 --> 55:42.820 And certainly also looking at certain faults that can occur there and 55:42.820 --> 55:48.640 this is another part there but we'll just look at this capability to 55:48.640 --> 55:51.140 decide on the routing of messages. 55:51.980 --> 55:55.420 The first thing we need for that is a standard for addresses. 55:55.940 --> 56:01.160 So IP, when it was designed initially, was designed with 32-bit 56:01.160 --> 56:09.320 addresses that means 4 bytes 4 bytes means 4 numbers between 0 and 56:09.320 --> 56:10.320 255. 56:10.960 --> 56:15.620 Not all the numbers are available, certain are used for different 56:15.620 --> 56:16.080 purposes. 56:16.860 --> 56:22.200 So you have always these numbers and these are the addresses of hosts, 56:22.220 --> 56:27.900 as you know and this is not just the address of a computer, a specific 56:27.900 --> 56:33.280 computer, but it is split up into an address of a network and the 56:33.280 --> 56:36.380 address of a specific host in that network. 56:36.920 --> 56:39.560 And then there are different classes of addresses in the initial 56:39.560 --> 56:44.460 version, meanwhile we are beyond that and have a more general way of 56:44.460 --> 56:46.860 stating that but just for information. 56:47.640 --> 56:53.560 The different classes were A, B, C, D and E classes so the differences 56:53.560 --> 57:00.760 were that you had different split ups between network address and host 57:00.760 --> 57:01.260 address. 57:01.860 --> 57:08.200 So in the A class networks you had only very few networks, just 7 bits 57:08.200 --> 57:16.780 for specifying the network and then the remaining 24 bits here for 57:16.780 --> 57:17.800 specifying the host. 57:18.340 --> 57:22.420 And this went down to the C class where you had many different 57:22.420 --> 57:28.140 networks that you could address but only here 8 bits for specifying 57:28.140 --> 57:28.660 the host. 57:29.920 --> 57:34.700 Fixed separations between network address and host address. 57:35.460 --> 57:40.360 Then you had a special class, so called multicast addresses where 57:40.360 --> 57:47.000 messages were indicated that they should be sent to all the nodes, not 57:47.000 --> 57:51.140 just to a specific host but to all the nodes and there was another 57:51.140 --> 57:53.120 class for so called future use. 57:53.620 --> 57:58.100 Meanwhile we definitely use all these in a different way. 57:58.180 --> 58:03.020 Actually we use it in a way, meanwhile that we have two 32 bit 58:03.020 --> 58:10.760 addresses and the second is just some kind of pattern of zeros and 58:10.760 --> 58:17.060 ones indicating it's some kind of filtering indicating the locations 58:17.060 --> 58:20.060 where the network is specified and the locations where the host is 58:20.060 --> 58:20.440 specified. 58:21.140 --> 58:28.040 So you have some very general way of actually specifying the way you 58:28.040 --> 58:29.400 address a certain computer. 58:30.540 --> 58:35.500 But this is the original scheme and the classes are indicated by the 58:35.500 --> 58:40.480 prefix of the number of ones here in the prefix of the addresses. 58:40.900 --> 58:46.380 That was the original scheme and you can immediately imagine if you 58:46.380 --> 58:51.160 look at that, it means that there are at most 2 to the 32 bit 58:51.160 --> 58:58.100 different addresses available and if you look at the number of 58:58.100 --> 59:02.720 computers that we have, the number of hosts well it's not that much so 59:02.720 --> 59:11.200 it's just 2 to the 32 is something like billions of computers we are 59:11.200 --> 59:16.200 beyond that meanwhile and so somehow this cannot be sufficient and so 59:16.200 --> 59:17.980 there is an upgrade of that. 59:18.180 --> 59:25.420 This is actually the so called IP version 4 addressing scheme. 59:25.640 --> 59:31.800 Meanwhile we have available the IP version 6 network addresses. 59:32.320 --> 59:37.040 I will come to that a bit later in the next time probably where I show 59:37.040 --> 59:38.740 you something about IP version 6. 59:38.960 --> 59:40.900 There we have 128 bit addresses. 59:41.220 --> 59:44.580 So much larger repository for addresses. 59:45.420 --> 59:50.080 So this is just mentioning again different characteristics of the 59:50.080 --> 59:56.580 networks and then what's happening locally if you look like this 59:56.580 --> 01:00:01.760 address is or has to be transformed into a physical address of a 01:00:01.760 --> 01:00:06.880 computer into an Ethernet address and this is then a physical address 01:00:06.880 --> 01:00:08.300 which is 01:00:13.080 --> 01:00:16.880 retrieved by transforming the IP address into a physical address that 01:00:16.880 --> 01:00:21.780 is recorded locally, done locally by the address resolution protocol. 01:00:22.240 --> 01:00:26.100 And then we have the physical address, how the host actually is 01:00:26.100 --> 01:00:31.060 addressed in the local area network. 01:00:32.160 --> 01:00:38.280 So now we have the classless inter-domain routing scheme, which is the 01:00:38.280 --> 01:00:41.080 generalization of those five different classes. 01:00:41.800 --> 01:00:46.500 And we also have the IP addressing scheme in IP version six, which is 01:00:46.500 --> 01:00:53.360 more and more now putting into practice or putting into real life. 01:00:53.600 --> 01:00:59.120 It has been around for more than 10 years now, but the process of 01:00:59.120 --> 01:01:01.700 establishing that was very, very, very slow. 01:01:03.260 --> 01:01:08.040 Okay, now if you have just these numbers as addresses, you cannot 01:01:08.040 --> 01:01:09.440 really memorize them. 01:01:09.460 --> 01:01:09.880 I cannot. 01:01:09.880 --> 01:01:14.760 I never know the number of my computer, but I certainly know the name 01:01:14.760 --> 01:01:15.560 of my computer. 01:01:15.840 --> 01:01:19.000 So we have the domain name system. 01:01:19.700 --> 01:01:23.660 And in that domain name system, as you know, we have certain naming 01:01:23.660 --> 01:01:24.280 conventions. 01:01:25.020 --> 01:01:31.220 So here, if I look at aifb.kit, oops, dot not D-E, what is that? 01:01:32.380 --> 01:01:33.360 E-D-U. 01:01:37.880 --> 01:01:39.460 E-D-U. 01:01:41.420 --> 01:01:45.360 Well, now it's visible, E-D-U. 01:01:45.700 --> 01:01:50.620 So corresponds to a certain URL, a certain IP address. 01:01:51.240 --> 01:01:57.060 And to know that aifb is the local, the most local part of that 01:01:57.060 --> 01:01:59.300 domain, kit.edu is a bit larger. 01:01:59.640 --> 01:02:03.980 E-D-U certainly is a larger network of education sites. 01:02:04.620 --> 01:02:12.060 So we have different domains that are established there initially. 01:02:12.660 --> 01:02:16.640 It was just the educational institutions, commercial institutions, 01:02:17.000 --> 01:02:19.520 government institutions, and some organizations. 01:02:20.020 --> 01:02:25.640 We have the country-specific domain names, and we have now all kinds 01:02:25.640 --> 01:02:29.060 of other domain names which are around, which have been established in 01:02:29.060 --> 01:02:29.760 recent years. 01:02:31.240 --> 01:02:37.960 And it certainly is important to be able to use those specific names 01:02:40.770 --> 01:02:43.200 and not have to use always the numbers. 01:02:43.500 --> 01:02:50.780 The only problem is in order to be able to communicate, you need those 01:02:50.780 --> 01:02:51.240 numbers. 01:02:52.200 --> 01:02:56.360 And so in order to actually send information from one location to the 01:02:56.360 --> 01:03:01.880 other, you have to contact the name server, which is providing you 01:03:01.880 --> 01:03:07.060 with the original IP address corresponding to that domain name. 01:03:07.860 --> 01:03:12.020 And if the domain name server is not working, then you cannot 01:03:12.020 --> 01:03:12.480 communicate. 01:03:13.040 --> 01:03:16.380 But if you still know the number, you can communicate. 01:03:16.600 --> 01:03:20.020 So it's quite often there are problems in communication between, or 01:03:20.020 --> 01:03:23.680 because domain name servers are out of service, but if you know the 01:03:23.680 --> 01:03:24.820 numbers, you can still communicate. 01:03:24.820 --> 01:03:29.460 Okay, so this is just looking at the domain name system, very 01:03:29.460 --> 01:03:33.480 important for ease of use of those addresses. 01:03:34.120 --> 01:03:37.880 Initially, when the networks were established, there were several 01:03:37.880 --> 01:03:41.380 different naming schemes and it was very complicated to actually send 01:03:41.380 --> 01:03:44.320 information from one location to another because it needed sequences 01:03:44.320 --> 01:03:48.720 of addresses which had to be arranged in an appropriate way. 01:03:49.960 --> 01:03:57.320 And yeah, what you know is that you can purchase domain names and then 01:03:57.320 --> 01:04:02.440 you have your own domain for your company or your personal use. 01:04:03.220 --> 01:04:07.420 Okay, those were the addresses, the domains, the domain name system. 01:04:07.820 --> 01:04:13.060 And now we come to what I indicated as the header that has to be 01:04:13.060 --> 01:04:19.000 provided by a certain protocol in order to provide the information for 01:04:19.000 --> 01:04:22.180 the receiving side that it knows what to do with that datagram. 01:04:22.420 --> 01:04:22.960 You have a question? 01:04:30.420 --> 01:04:35.520 There will be more domain, so the number of domain or the domain names 01:04:35.520 --> 01:04:40.360 will always be extended by new names. 01:04:40.420 --> 01:04:45.200 As I indicated here, there are some names down there and they are 01:04:45.200 --> 01:04:56.140 managed by those organizations like InterNIC and other organizations 01:04:56.140 --> 01:04:57.160 also in Germany. 01:04:59.680 --> 01:05:02.960 I don't actually understand your question. 01:05:36.590 --> 01:05:37.090 Well, if... 01:05:41.290 --> 01:05:47.270 Yes, so the question was about the different or new developments were 01:05:47.270 --> 01:05:53.230 proprietary domain, like top-level domain names actually were or can 01:05:53.230 --> 01:05:53.810 be registered. 01:05:54.650 --> 01:05:59.430 This certainly makes the management of the systems more complex, but I 01:05:59.430 --> 01:06:02.810 think we have sufficiently powerful systems in the world to deal with 01:06:02.810 --> 01:06:03.090 that. 01:06:04.150 --> 01:06:10.210 It is something which is providing more convenience for the users. 01:06:10.490 --> 01:06:15.990 It makes it like you can have your own domain and then this still has 01:06:15.990 --> 01:06:22.010 to be mapped appropriately to the actual numbers. 01:06:22.810 --> 01:06:27.290 But it's a requirement on the domain name service that they actually 01:06:27.290 --> 01:06:34.890 can build up those large repositories of domain names and actually 01:06:34.890 --> 01:06:42.390 translate the domain names into the numbering systems, either in IP 01:06:42.390 --> 01:06:48.510 version 4 numbers or now in the IP version 6 numbers with 128 bits. 01:06:49.190 --> 01:06:54.590 So this is just making the task of managing the domain names a bit 01:06:54.590 --> 01:07:02.550 more complex, but since our computers are getting more powerful, it is 01:07:02.550 --> 01:07:03.550 feasible to do that. 01:07:03.850 --> 01:07:04.130 Yeah? 01:07:07.240 --> 01:07:11.920 So what kind of information is actually necessary if we look at the IP 01:07:11.920 --> 01:07:12.640 protocol? 01:07:12.860 --> 01:07:15.080 Like what does it have to do? 01:07:15.600 --> 01:07:22.020 I said we have a certain datagram where actually there is already a 01:07:22.020 --> 01:07:24.160 header there from TCP. 01:07:26.460 --> 01:07:28.300 And then we have here certain data. 01:07:29.900 --> 01:07:34.680 And now we put something in front, which is the IP header. 01:07:35.260 --> 01:07:41.300 The IP header is a sequence of, certainly a sequence of 0s and 1s, a 01:07:41.300 --> 01:07:42.000 sequence of bytes. 01:07:42.480 --> 01:07:44.860 This is organized into 32-bit words. 01:07:45.160 --> 01:07:49.220 So 32-bit, we know that for a long time we had 32-bit computers. 01:07:49.900 --> 01:07:54.920 So 32-bit is the word size built from, and it's also the address size 01:07:54.920 --> 01:07:58.180 of the IP version 4. 01:07:58.800 --> 01:08:00.440 So we have 32 bits here. 01:08:00.940 --> 01:08:10.360 That means that we have here, as you can see, actually 1, 2, 3, 4, 5 01:08:10.360 --> 01:08:15.380 essential words that have to be in every IP header. 01:08:16.380 --> 01:08:19.000 After that, we have certain optional parts. 01:08:19.000 --> 01:08:23.460 And so the IP header has 20 to 60 bytes. 01:08:25.140 --> 01:08:27.920 So it has at least 20, it may have 60 bytes. 01:08:28.480 --> 01:08:31.200 And now you see all kinds of information on that. 01:08:31.460 --> 01:08:34.240 Version, length, service type, total length, and so on. 01:08:34.860 --> 01:08:42.080 And on, well, here's some more text which I animated in the wrong 01:08:42.080 --> 01:08:42.460 sequence. 01:08:42.800 --> 01:08:45.000 So that's what I just told you. 01:08:45.300 --> 01:08:46.900 That is the structure of the IP datagram. 01:08:47.280 --> 01:08:49.940 And I would like to... Here's the datagram again. 01:08:50.400 --> 01:08:53.640 And I will give you more information on the individual things, why 01:08:53.640 --> 01:08:54.440 they are in there. 01:08:55.140 --> 01:08:58.580 The first thing is that since there are different versions of the IP 01:08:58.580 --> 01:09:03.240 protocol, you have to indicate the version number of the protocol so 01:09:03.240 --> 01:09:07.340 that the receiving side knows how to process that datagram. 01:09:09.300 --> 01:09:11.980 Then you need information on the length. 01:09:11.980 --> 01:09:21.080 I said that the number of words in the header can vary between... 01:09:21.080 --> 01:09:23.980 or the number of bytes between 20 and 60. 01:09:24.520 --> 01:09:28.560 And so you have to specify the length of the header. 01:09:29.700 --> 01:09:34.660 Because then the receiving side knows how much information actually 01:09:34.660 --> 01:09:39.540 has to be looked at and what is the next level information. 01:09:43.040 --> 01:09:45.720 Certainly like 60 bytes is not that much. 01:09:46.140 --> 01:09:48.880 The option part has maximally 40 bytes. 01:09:49.460 --> 01:09:51.080 That is not very much information. 01:09:51.260 --> 01:09:53.360 And there are many things that have to be recorded there. 01:09:53.460 --> 01:09:58.080 For example, different locations that are passed through on the path 01:09:58.080 --> 01:09:59.440 from source to destination. 01:10:00.200 --> 01:10:03.040 There's not that much room for specifying a lot of information there. 01:10:03.180 --> 01:10:07.400 So this is a restriction that was completely sufficient in the early 01:10:07.400 --> 01:10:09.440 days of the Internet, but not nowadays. 01:10:10.200 --> 01:10:11.880 Then the service type. 01:10:12.160 --> 01:10:15.720 This is something which was designed originally, in the original 01:10:15.720 --> 01:10:18.040 version of the IP protocol. 01:10:18.580 --> 01:10:20.820 So it desired quality of service. 01:10:21.340 --> 01:10:23.820 But actually it was never used. 01:10:24.120 --> 01:10:28.920 So the original idea was to look at, for example, priority, 01:10:29.120 --> 01:10:31.700 reliability, throughput, delay. 01:10:32.420 --> 01:10:37.620 You could specify different aspects that should be looked at if you 01:10:37.620 --> 01:10:42.500 decide on how to forward the packet. 01:10:42.700 --> 01:10:44.680 As I said, we have a certain node. 01:10:44.900 --> 01:10:46.460 It gets some message. 01:10:46.920 --> 01:10:50.460 And then there are different possibilities to forward it to next hops. 01:10:51.080 --> 01:10:54.780 Now if you know something about the reliability of those different 01:10:54.780 --> 01:11:02.920 next hops, or next routers, you can decide on the most reliable one. 01:11:03.560 --> 01:11:04.840 Maybe it's not the fastest. 01:11:04.840 --> 01:11:09.620 But if reliability is the major point, then you will use the most 01:11:09.620 --> 01:11:10.380 reliable one. 01:11:11.320 --> 01:11:15.360 You may have some other priority, or you may have throughput, or 01:11:15.360 --> 01:11:15.820 delay. 01:11:16.240 --> 01:11:22.420 So these can be used essentially in the... 01:11:22.420 --> 01:11:24.180 or can be specified in the IP protocol. 01:11:24.680 --> 01:11:33.360 But the protocol usually is not actually taking notice of that. 01:11:33.360 --> 01:11:36.160 So this is something which was there, but not really put into 01:11:36.160 --> 01:11:36.660 practice. 01:11:37.600 --> 01:11:39.080 And then you need more information. 01:11:39.200 --> 01:11:41.580 You need information about the length of the datagram. 01:11:41.900 --> 01:11:48.780 How many bytes are there in the datagram in total? 01:11:49.700 --> 01:11:53.220 Because you have to know how much information you have to look at. 01:11:53.480 --> 01:11:58.220 How much information you have to move on to the next hop, or move on 01:11:58.220 --> 01:11:59.940 to the next layer, to the TCP layer. 01:12:00.820 --> 01:12:05.940 The maximum here is since this is like the length, the total length is 01:12:05.940 --> 01:12:07.960 16 bits. 01:12:08.420 --> 01:12:11.840 So you have 2 to the 16 different bytes. 01:12:13.260 --> 01:12:16.380 And this is the maximum size of the datagram. 01:12:18.000 --> 01:12:22.760 You can have... like I stated here, but this is not really sufficient 01:12:22.760 --> 01:12:23.840 for gigabit networks. 01:12:24.040 --> 01:12:25.440 Why not for gigabit networks? 01:12:25.440 --> 01:12:32.300 Because there you have a lot of information that is sent, and you 01:12:32.300 --> 01:12:33.440 always should have... 01:12:34.460 --> 01:12:39.980 or you should try to reduce the administrative overhead that you have 01:12:39.980 --> 01:12:41.900 to spend for sending information. 01:12:42.600 --> 01:12:49.600 And if you have gigabit networks, then you have to really operate very 01:12:49.600 --> 01:12:52.020 often on those administrative data. 01:12:52.020 --> 01:13:03.500 And it would be more efficient to have larger jumbo datagrams, which 01:13:03.500 --> 01:13:05.600 are larger than those 64 kilobytes. 01:13:06.200 --> 01:13:08.740 So this is large datagrams. 01:13:09.200 --> 01:13:15.020 The normal size of datagrams is rather in the range of 1,500 bytes, or 01:13:15.020 --> 01:13:18.120 just a few thousand bytes. 01:13:19.860 --> 01:13:23.160 Then we have a field which is called identification. 01:13:23.740 --> 01:13:25.000 Why do we need the identification? 01:13:25.820 --> 01:13:27.780 It puts a nice name to the datagram. 01:13:28.100 --> 01:13:32.120 Now this is important, because sometimes you have... 01:13:32.940 --> 01:13:36.880 like maybe that you would like to send data over a certain number of 01:13:36.880 --> 01:13:41.580 hosts, and all of a sudden you notice, well, the next link is not 01:13:41.580 --> 01:13:44.040 capable of accepting large datagrams. 01:13:44.060 --> 01:13:45.400 You have to split the datagram. 01:13:45.880 --> 01:13:49.480 Like we have here one datagram having here all these headers in front. 01:13:50.040 --> 01:13:53.000 Now you have to split that into fragments. 01:13:53.360 --> 01:13:54.540 Let's say three fragments. 01:13:56.120 --> 01:14:01.560 Now fragments means you just chop that datagram into three pieces and 01:14:01.560 --> 01:14:05.620 provide them with different IP headers and send them off. 01:14:06.060 --> 01:14:09.600 Only the first one has the TCP header on it. 01:14:09.840 --> 01:14:10.980 The next ones are just data. 01:14:12.740 --> 01:14:17.320 Now you have to know at the receiving site that you have to reassemble 01:14:17.320 --> 01:14:20.100 those fragments into one original datagram. 01:14:21.060 --> 01:14:22.860 And for that you need the identification. 01:14:23.420 --> 01:14:28.960 So for that you need a name for the datagram, so all the fragments 01:14:28.960 --> 01:14:32.940 have the same name, and then you know how to reassemble the fragments 01:14:32.940 --> 01:14:34.980 into one datagram. 01:14:36.520 --> 01:14:37.820 So that's the identification. 01:14:38.780 --> 01:14:47.800 Certainly you also have to know something about the position of the 01:14:47.800 --> 01:14:53.920 fragments in the datagram because the sequence at which you actually 01:14:53.920 --> 01:14:58.860 receive datagrams or fragments may be different from the sequence at 01:14:58.860 --> 01:14:59.980 which they have been sent out. 01:15:01.500 --> 01:15:04.420 And so several things have to be looked at. 01:15:04.840 --> 01:15:09.080 There's one bit that can be specified here, one flag that can indicate 01:15:09.080 --> 01:15:12.900 that you should not fragment the datagram. 01:15:13.320 --> 01:15:15.760 Things like that might be necessary sometimes. 01:15:16.460 --> 01:15:23.420 So if you have this flag, if this is set, then the IP router must not 01:15:23.420 --> 01:15:37.790 actually continue. 01:15:39.050 --> 01:15:42.290 So you can specify that you don't want to fragment a datagram. 01:15:42.510 --> 01:15:46.710 You also have to indicate whether there will be more fragments. 01:15:47.450 --> 01:15:52.050 So only the last fragment will not have that flag set. 01:15:52.050 --> 01:15:55.710 Because it's the last one, so in this way you can find out what is the 01:15:55.710 --> 01:15:59.810 last fragment actually, at what point in time you actually finished 01:15:59.810 --> 01:16:00.630 with your datagram. 01:16:01.030 --> 01:16:06.630 So the first two here in this example would have this M field set. 01:16:06.990 --> 01:16:09.010 The last one would not have it set. 01:16:09.610 --> 01:16:11.910 Like the headers for those fragments. 01:16:12.910 --> 01:16:16.910 And then there is the fragment offset, which indicates the position of 01:16:16.910 --> 01:16:22.450 the fragment in the datagram in multiples of 8 bytes. 01:16:23.650 --> 01:16:30.830 And so this is just certainly important because you have to know what 01:16:30.830 --> 01:16:36.670 are the positions where a specific datagram has to be put, or where a 01:16:36.670 --> 01:16:40.230 fragment has to be put in order to reconstruct the original datagram. 01:16:40.690 --> 01:16:44.110 So this is fragmentation and how you can deal with fragmentation. 01:16:44.770 --> 01:16:48.930 You need this information in order to be able to actually deal 01:16:48.930 --> 01:16:51.990 appropriately with such a fragmented system. 01:16:53.370 --> 01:16:57.230 And then another point is, another information which is necessary, 01:16:57.430 --> 01:17:01.190 which has to be provided, is the time-to-live indication. 01:17:01.430 --> 01:17:07.170 Time-to-live means that we don't want a situation occurring where a 01:17:07.170 --> 01:17:11.490 datagram is actually moving around in cycles in the network because 01:17:11.490 --> 01:17:14.850 that would lead to an overflow of packets in the network. 01:17:14.850 --> 01:17:19.270 So we have to make sure that a datagram will be discarded after a 01:17:19.270 --> 01:17:19.990 number of hops. 01:17:22.250 --> 01:17:27.590 So this could be in time in seconds, but usually that counts the 01:17:27.590 --> 01:17:28.210 number of hops. 01:17:28.630 --> 01:17:34.930 Usually it's just set to an initial value in the beginning, and then 01:17:34.930 --> 01:17:37.250 it's decremented at every hop. 01:17:37.750 --> 01:17:42.650 And so as soon as it is down to zero, that packet is discarded and not 01:17:42.650 --> 01:17:43.370 forwarded further. 01:17:43.370 --> 01:17:48.870 So it shows that this is a way of automatic garbage collection. 01:17:49.950 --> 01:17:54.890 So old packets just get discarded automatically because their time-to 01:17:54.890 --> 01:17:57.890 -live field will be set to zero at some point. 01:17:58.870 --> 01:18:04.610 And so this is necessary for that purpose for garbage collection. 01:18:05.330 --> 01:18:07.010 Then we have another thing. 01:18:07.070 --> 01:18:07.830 We have the protocol. 01:18:08.830 --> 01:18:14.930 So we have to specify which type or which protocol should be used on 01:18:14.930 --> 01:18:16.190 the next layer. 01:18:16.370 --> 01:18:21.230 So if you are at the destination, it has to be indicated whether the 01:18:21.230 --> 01:18:24.730 next layer protocol, the transport layer, is TCP or UDP. 01:18:25.970 --> 01:18:30.030 UDP is the user datagram protocol which treats the datagrams 01:18:30.030 --> 01:18:33.550 differently from the transport or from the transmission control 01:18:33.550 --> 01:18:33.970 protocol. 01:18:33.970 --> 01:18:40.590 And so this has to be indicated by the next entry there, the protocol. 01:18:41.450 --> 01:18:44.130 Then we have the header checksum. 01:18:44.670 --> 01:18:50.950 This is just providing information on whether some bit errors have 01:18:50.950 --> 01:18:58.290 occurred, for example, on the transport, like while it was moved 01:18:58.290 --> 01:18:58.650 somewhere. 01:18:58.650 --> 01:19:07.110 And so you just look at the sum of all the 16-bit half words in the 01:19:07.110 --> 01:19:12.650 header without this field, and then you use the ones complement of 01:19:12.650 --> 01:19:13.230 that sum. 01:19:13.510 --> 01:19:18.690 And so if you add all the 16-bit fields of the header, the result 01:19:18.690 --> 01:19:25.270 should be zero because the ones complement of the sum of all the other 01:19:25.270 --> 01:19:31.130 fields, if that is added to the sum of all the fields, then you get 01:19:31.130 --> 01:19:37.090 zero upon arrival, or there should be zero upon arrival. 01:19:37.250 --> 01:19:41.010 If you don't get zero as the result of adding up all the 16-bit 01:19:41.010 --> 01:19:45.170 fields, then some error has occurred and you have to discard that 01:19:45.170 --> 01:19:45.510 datagram. 01:19:46.850 --> 01:19:48.370 So this is the header checksum. 01:19:48.970 --> 01:19:50.810 Then there are the addresses. 01:19:51.130 --> 01:19:56.490 You have to know, well, where this datagram actually comes from. 01:19:56.590 --> 01:19:59.490 That's the source address and also the destination address. 01:20:00.630 --> 01:20:04.550 The source address is necessary because some information also may have 01:20:04.550 --> 01:20:12.230 to be communicated back to the source but this is dealt with a 01:20:12.230 --> 01:20:17.910 different protocol, the message control protocol, which will indicate 01:20:17.910 --> 01:20:25.890 the occurrence of certain failures to the source host. 01:20:26.530 --> 01:20:31.730 Then certainly you need the destination address because you have to 01:20:31.730 --> 01:20:35.650 know where to send your datagram to. 01:20:37.470 --> 01:20:42.410 These are the mandatory fields of the header. 01:20:43.130 --> 01:20:48.170 Then you have certain optional parts which may be there but need not 01:20:48.170 --> 01:20:48.730 be there. 01:20:48.730 --> 01:20:51.070 And these are for different purposes. 01:20:51.370 --> 01:20:55.010 For example, for security, you can specify the security level of the 01:20:55.010 --> 01:20:55.390 datagram. 01:20:56.630 --> 01:21:02.530 You cannot really use their encryption or like security is not 01:21:02.530 --> 01:21:06.150 supported in a way that we would nowadays think it's necessary. 01:21:06.410 --> 01:21:11.770 So in the IP version 6, this has changed and security is supported to 01:21:11.770 --> 01:21:13.830 a much larger extent. 01:21:14.510 --> 01:21:20.710 And then an important point is to specify the routing of the datagram. 01:21:21.250 --> 01:21:25.170 Strict routing means it specifies completely the path to follow. 01:21:25.610 --> 01:21:30.370 So it does not leave any options to the routers to decide what the 01:21:30.370 --> 01:21:31.530 next hop should be. 01:21:31.850 --> 01:21:33.350 But here you have strict routing. 01:21:33.490 --> 01:21:37.850 It indicates what is actually the sequence of nodes that have to be 01:21:37.850 --> 01:21:38.690 used there. 01:21:39.310 --> 01:21:42.970 You can also have some loose routing, specify just some routers on the 01:21:42.970 --> 01:21:43.330 path. 01:21:43.990 --> 01:21:48.470 Another thing would be to record the route of your packet, so then you 01:21:48.470 --> 01:21:54.310 can look at the trace later on and notice where your packet actually 01:21:54.310 --> 01:21:54.810 has been. 01:21:55.410 --> 01:22:01.690 This certainly is important for error analysis, for fault analysis. 01:22:02.770 --> 01:22:06.830 You may put in some timestamps, then you know at what time the 01:22:06.830 --> 01:22:08.970 datagram has been at certain locations. 01:22:09.510 --> 01:22:16.450 But all this has to fit into the 40 extra bytes that are available in 01:22:16.450 --> 01:22:17.310 those option paths. 01:22:19.810 --> 01:22:25.670 So the length of the option field is always a multiple of four bytes. 01:22:27.050 --> 01:22:33.190 And so if the options are ending somewhere before that, you have to 01:22:33.190 --> 01:22:36.130 fill up those with additional bits. 01:22:36.930 --> 01:22:39.010 And then this is the end of the header. 01:22:39.290 --> 01:22:41.330 And after that, the data is coming. 01:22:41.330 --> 01:22:47.650 The data may be original data of some fragment of the datagram, or it 01:22:47.650 --> 01:22:52.850 may be just the next header for the TCP protocol. 01:22:54.150 --> 01:22:54.270 Okay. 01:22:55.470 --> 01:22:58.530 So this is all the different fields. 01:22:58.830 --> 01:23:02.050 And as soon as you would remove one of those fields, certain 01:23:02.050 --> 01:23:04.750 functionality could not be provided by the IP protocol. 01:23:05.470 --> 01:23:11.490 So they are all necessary to provide security, like a header checksum 01:23:11.490 --> 01:23:13.510 or fault detection. 01:23:14.150 --> 01:23:15.430 Provide connectivity. 01:23:15.850 --> 01:23:16.650 You need the addresses. 01:23:17.450 --> 01:23:19.130 You need information on the size. 01:23:19.330 --> 01:23:22.450 You need information on whether you can fragment the datagram. 01:23:23.210 --> 01:23:28.310 And you have to make sure that you can get garbage collection. 01:23:28.430 --> 01:23:31.870 The garbage collection is dealt with in a very elegant way by just 01:23:31.870 --> 01:23:35.130 counting down to zero, and then you have... 01:23:35.130 --> 01:23:38.430 or you just discard such a message. 01:23:39.290 --> 01:23:39.970 Okay. 01:23:40.670 --> 01:23:44.950 And now we have to look at how we actually use the information that we 01:23:44.950 --> 01:23:49.150 have in the IP header for routing. 01:23:49.850 --> 01:23:56.830 So this is our node, or router, gets a certain datagram and has to 01:23:56.830 --> 01:24:01.250 decide where to forward or what is the next hop, actually. 01:24:02.130 --> 01:24:07.370 So what the router actually does is it performs a certain router 01:24:07.370 --> 01:24:11.470 algorithm which here is indicated in a very abstract way. 01:24:12.090 --> 01:24:15.250 It's certainly not exactly the routing algorithm that is used, but 01:24:15.250 --> 01:24:16.070 similar to that. 01:24:16.670 --> 01:24:18.730 So what you need is... 01:24:18.730 --> 01:24:20.030 you look at a datagram. 01:24:20.790 --> 01:24:23.690 And then you have to look at the destination address. 01:24:23.910 --> 01:24:27.390 You have to extract the destination address from the datagram from the 01:24:27.390 --> 01:24:29.330 header, from the IP header. 01:24:31.350 --> 01:24:37.630 Then you have the destination address From the destination address, 01:24:38.090 --> 01:24:41.450 you first of all get the network address. 01:24:41.790 --> 01:24:45.710 I told you that we have the part specifying the network address and 01:24:45.710 --> 01:24:46.510 the host address. 01:24:46.870 --> 01:24:49.950 First of all, you have to know to which network you have to send it 01:24:49.950 --> 01:24:50.090 to. 01:24:51.330 --> 01:24:57.410 And then, if that network address is in the list of direct network 01:24:57.410 --> 01:25:02.690 addresses of that host, so if this router here is in some way 01:25:02.690 --> 01:25:08.010 connected to some local area network and you have a direct network 01:25:08.010 --> 01:25:13.790 address there, then you send the datagram to the physical address of 01:25:13.790 --> 01:25:16.530 that host that is specified there. 01:25:16.770 --> 01:25:18.370 So that is the destination address. 01:25:18.930 --> 01:25:22.130 This destination address is transformed into the physical address. 01:25:22.670 --> 01:25:24.170 For example, an Ethernet address. 01:25:24.970 --> 01:25:29.030 And this is then sent locally to the host. 01:25:29.630 --> 01:25:33.950 If it's not in the direct network addresses, then it's some other kind 01:25:33.950 --> 01:25:34.470 of address. 01:25:35.370 --> 01:25:42.030 And then you look at whether the option field has some strict routing 01:25:42.030 --> 01:25:43.190 information. 01:25:43.330 --> 01:25:49.870 If it's strict routing, then you have to send the datagram to the next 01:25:49.870 --> 01:25:55.730 hop that is specified in the strict routing option that's indicated 01:25:55.730 --> 01:25:58.710 here as datagram.option.myroute. 01:25:59.650 --> 01:26:03.050 Some next hop that has to be used there. 01:26:03.810 --> 01:26:07.510 If this is not the case, then you just look it up in your routing 01:26:07.510 --> 01:26:08.030 table. 01:26:08.330 --> 01:26:11.790 You have all kinds of different destinations. 01:26:12.650 --> 01:26:13.710 And you have next links. 01:26:13.810 --> 01:26:18.710 So if the destination is certain, is here some location A, then you 01:26:18.710 --> 01:26:21.370 get to the next link L, next hop. 01:26:21.910 --> 01:26:25.550 And this somehow has to be specified in this routing table. 01:26:26.750 --> 01:26:33.470 And if this network address is not in the routing table, then still 01:26:33.470 --> 01:26:34.550 you have to do something. 01:26:35.010 --> 01:26:38.450 So then you just send the datagram to some default router, and that 01:26:38.450 --> 01:26:42.190 default router has to take care of the further processing of the 01:26:42.190 --> 01:26:42.490 datagram. 01:26:42.670 --> 01:26:44.130 So this is just to make it complete. 01:26:44.850 --> 01:26:47.570 And if all this does not work, then you have to report an error. 01:26:47.930 --> 01:26:51.490 So this is just showing an abstract view of such a routing algorithm. 01:26:51.970 --> 01:26:55.590 And the essential part in there is this routing table. 01:26:56.110 --> 01:27:00.650 And next week we will look at the way we actually set up those routing 01:27:00.650 --> 01:27:02.750 tables by different algorithms. 01:27:03.530 --> 01:27:04.790 Okay, that's it for today. 01:27:04.990 --> 01:27:05.310 Thank you.