WEBVTT 1 00:00:01.439 --> 00:00:03.439 - Join us as we gather around the hedge, 2 00:00:03.439 --> 00:00:04.819 - where we dig into technology, 3 00:00:05.120 --> 00:00:07.919 - business, and culture with the finest minds in 4 00:00:07.919 --> 00:00:08.820 - computer networking. 5 00:00:21.054 --> 00:00:23.954 - Well, good morning, Tom, and good morning, Audrey. 6 00:00:24.850 --> 00:00:26.630 - Good morning. Audrey says good morning. 7 00:00:28.130 --> 00:00:30.530 - Audrey is the plant for those who aren't 8 00:00:30.530 --> 00:00:31.350 - paying attention. 9 00:00:31.650 --> 00:00:33.729 - But don't don't don't listen to a lot 10 00:00:33.729 --> 00:00:34.390 - of edge. 11 00:00:34.690 --> 00:00:35.190 - Yeah. 12 00:00:38.524 --> 00:00:40.545 - So how is the weather where you are? 13 00:00:40.844 --> 00:00:43.964 - Weather's great. It is mild. It is not 14 00:00:43.964 --> 00:00:47.344 - too hot. Pretty great for, April in Texas. 15 00:00:47.405 --> 00:00:47.725 - So 16 00:00:48.284 --> 00:00:50.524 - Okay. Well, there you go. Yeah. I'm just 17 00:00:50.524 --> 00:00:52.225 - back from Chicago, which is, 18 00:00:52.870 --> 00:00:54.729 - anyway, we won't talk about Chicago. 19 00:00:58.229 --> 00:00:58.729 - So 20 00:00:59.350 --> 00:01:01.510 - we are joined this morning by t h 21 00:01:01.510 --> 00:01:02.950 - Thomas. I don't know how to say your 22 00:01:02.950 --> 00:01:04.310 - last name, so I'm just not even gonna 23 00:01:04.469 --> 00:01:06.150 - I'm not gonna do it. I'm I'm gonna 24 00:01:06.150 --> 00:01:08.884 - get myself in trouble, and it's better. It's 25 00:01:08.884 --> 00:01:09.825 - all good. 26 00:01:12.685 --> 00:01:15.004 - So so where are you, T. H? What 27 00:01:15.004 --> 00:01:15.504 - what 28 00:01:16.125 --> 00:01:18.844 - Yeah. So hello everybody. So I am in, 29 00:01:19.084 --> 00:01:21.905 - France in Strasbourg. It's a it's a town 30 00:01:21.965 --> 00:01:24.224 - near Germany but on the French side. 31 00:01:24.540 --> 00:01:26.460 - And so the there is a university here, 32 00:01:26.460 --> 00:01:28.400 - and I am postdoc at this university. 33 00:01:29.659 --> 00:01:32.380 - Ah, interesting. So Strasbourg. Yeah. Yes. I always 34 00:01:32.380 --> 00:01:35.020 - think of Strasbourg as being in Germany for 35 00:01:35.020 --> 00:01:36.314 - some reason in my head, 36 00:01:36.795 --> 00:01:39.034 - but it's actually on the yeah. It's on 37 00:01:39.034 --> 00:01:41.135 - the French side of the border. But, like, 38 00:01:41.755 --> 00:01:42.415 - you know? 39 00:01:44.234 --> 00:01:46.954 - The the name sounds super super German indeed. 40 00:01:46.954 --> 00:01:47.855 - Yes. It 41 00:01:48.234 --> 00:01:48.975 - really does. 42 00:01:49.355 --> 00:01:51.194 - So we'll just say you're in Germany just 43 00:01:51.194 --> 00:01:53.520 - to, like, keep the move going. That'll be 44 00:01:54.939 --> 00:01:55.920 - simpler thing. 45 00:01:56.299 --> 00:01:58.460 - And then we have t eight Thomas Alfori. 46 00:01:58.460 --> 00:01:59.180 - Is that the right 47 00:02:00.299 --> 00:02:02.540 - Alfori, but Alfori will be will be fine 48 00:02:02.540 --> 00:02:04.799 - in today. Yeah. He's fine as well. Whatever. 49 00:02:05.995 --> 00:02:07.515 - And so where where Thomas, 50 00:02:08.715 --> 00:02:11.594 - also based in Strasbourg, France, which by the 51 00:02:11.594 --> 00:02:13.375 - way is the best city in the world. 52 00:02:14.474 --> 00:02:16.715 - And I'm also a postdoc at the University 53 00:02:16.715 --> 00:02:17.835 - of Strasbourg working, 54 00:02:18.155 --> 00:02:20.814 - on several stuffs with, Thomas on top of. 55 00:02:21.240 --> 00:02:23.819 - Okay. Awesome. That's really great. So postdocs, 56 00:02:24.439 --> 00:02:26.060 - which is which is really cool. 57 00:02:26.919 --> 00:02:27.419 - Alright. 58 00:02:27.879 --> 00:02:30.039 - So let's just begin at a simple place. 59 00:02:30.039 --> 00:02:33.159 - We're we're talking about bgproutes.io, 60 00:02:33.159 --> 00:02:34.699 - and let's just begin with 61 00:02:35.319 --> 00:02:37.575 - what problem are you trying to solve? Like, 62 00:02:37.575 --> 00:02:39.375 - what when you went into this, I am 63 00:02:39.375 --> 00:02:40.895 - assuming you did this as part of your 64 00:02:40.895 --> 00:02:41.395 - postdoc. 65 00:02:41.855 --> 00:02:43.635 - You know, this is like a postdoc project. 66 00:02:44.094 --> 00:02:46.335 - And so, you know, you're not doing it 67 00:02:46.335 --> 00:02:48.594 - for free, but close enough. And 68 00:02:50.400 --> 00:02:52.400 - that's that's the way post doc work usually 69 00:02:52.400 --> 00:02:53.300 - is. Yes. 70 00:02:55.040 --> 00:02:57.199 - So so what problem are you trying to 71 00:02:57.199 --> 00:02:57.699 - solve? 72 00:02:58.560 --> 00:03:01.040 - Yep. So it all started, indeed a few 73 00:03:01.040 --> 00:03:02.260 - years ago when, 74 00:03:02.955 --> 00:03:05.215 - TA did his PhD at the university 75 00:03:05.995 --> 00:03:06.655 - of Strasbourg 76 00:03:07.194 --> 00:03:07.514 - and, 77 00:03:08.394 --> 00:03:10.235 - you know, at that time we wanted to 78 00:03:10.235 --> 00:03:14.014 - better analyze Internet routing, so better detect hijacks, 79 00:03:14.235 --> 00:03:15.775 - better detect route leaks, 80 00:03:16.319 --> 00:03:16.819 - better, 81 00:03:17.680 --> 00:03:20.240 - analyze the Internet topology, you know, the ice 82 00:03:20.240 --> 00:03:21.060 - level topology. 83 00:03:21.680 --> 00:03:24.080 - And at that time, we actually realized that 84 00:03:24.080 --> 00:03:26.020 - if we want to do all of this 85 00:03:26.080 --> 00:03:29.040 - accurately, we need also accurate data. We need 86 00:03:29.040 --> 00:03:30.500 - as much data as possible. 87 00:03:31.284 --> 00:03:32.025 - And so 88 00:03:32.564 --> 00:03:34.724 - that's why we we really thought about how 89 00:03:34.724 --> 00:03:36.504 - we could collect more data. 90 00:03:37.205 --> 00:03:39.284 - And, so we did a lot of research 91 00:03:39.284 --> 00:03:41.205 - on that because it's not so simple. And 92 00:03:41.205 --> 00:03:43.044 - and a couple of years ago, we could 93 00:03:43.044 --> 00:03:44.264 - publish a paper 94 00:03:44.719 --> 00:03:46.659 - at a conference that is called CITCOM, 95 00:03:46.960 --> 00:03:49.459 - it's among the best conference, and 96 00:03:50.080 --> 00:03:52.900 - we realized that people really liked the idea 97 00:03:53.120 --> 00:03:55.379 - and many operators and researchers 98 00:03:55.759 --> 00:03:56.259 - motivated 99 00:03:57.280 --> 00:03:57.939 - us to 100 00:03:58.574 --> 00:04:00.335 - actually go to that direction and not just 101 00:04:00.335 --> 00:04:02.655 - having a research paper, but to actually build 102 00:04:02.655 --> 00:04:03.955 - a system like that, 103 00:04:04.495 --> 00:04:07.375 - so that it becomes really practical operational and 104 00:04:07.375 --> 00:04:09.534 - and it becomes like something in the real 105 00:04:09.534 --> 00:04:10.034 - life. 106 00:04:10.520 --> 00:04:13.080 - So that's what we yeah. Okay. So let's 107 00:04:13.080 --> 00:04:15.979 - let's I have a few questions there. So 108 00:04:16.120 --> 00:04:16.620 - traditionally 109 00:04:17.240 --> 00:04:18.060 - or historically, 110 00:04:18.759 --> 00:04:21.019 - whenever anybody wants to collect 111 00:04:22.600 --> 00:04:24.379 - data off the default reason, 112 00:04:25.354 --> 00:04:27.214 - you immediately run into, 113 00:04:28.314 --> 00:04:30.954 - no, but it's my data, and, no, I'm 114 00:04:30.954 --> 00:04:33.194 - not going to tell you I'm not going 115 00:04:33.194 --> 00:04:35.134 - to give you BGP peering information, 116 00:04:35.595 --> 00:04:37.615 - or I'm not gonna give you full routes, 117 00:04:37.834 --> 00:04:40.139 - or whatever the case might happen to be. 118 00:04:40.220 --> 00:04:42.000 - Right? That's that's a very 119 00:04:42.300 --> 00:04:44.879 - because it's a business it's a it's a 120 00:04:45.100 --> 00:04:45.600 - business 121 00:04:45.900 --> 00:04:46.400 - secret. 122 00:04:46.939 --> 00:04:49.660 - It's like the Coca Cola formula. It's held 123 00:04:49.660 --> 00:04:50.879 - in a in a 124 00:04:51.500 --> 00:04:52.720 - in some type of 125 00:04:53.254 --> 00:04:54.714 - of a safe someplace, 126 00:04:55.254 --> 00:04:56.794 - stuffed behind the CEO's 127 00:04:57.574 --> 00:05:00.134 - picture of Van Gogh or whatever they have 128 00:05:00.134 --> 00:05:02.475 - in their office. So yeah. 129 00:05:03.654 --> 00:05:05.014 - So did you find that to be a 130 00:05:05.014 --> 00:05:06.855 - problem? Is that, like, something that you ran 131 00:05:06.855 --> 00:05:07.355 - into? 132 00:05:08.319 --> 00:05:11.439 - Yeah. Definitely. So you are right that most 133 00:05:11.439 --> 00:05:12.339 - of the operators, 134 00:05:12.720 --> 00:05:14.500 - they are not willing 135 00:05:14.879 --> 00:05:16.100 - to share data. 136 00:05:17.120 --> 00:05:19.300 - In fact, because it's a sort of confidential, 137 00:05:20.080 --> 00:05:21.764 - we don't they don't want, for instance, 138 00:05:22.965 --> 00:05:24.165 - that we know their routing policies and with 139 00:05:24.165 --> 00:05:25.764 - who they are appearing and this kind of 140 00:05:25.764 --> 00:05:26.985 - things. However, 141 00:05:27.845 --> 00:05:28.345 - there 142 00:05:28.884 --> 00:05:31.845 - are still many operators that that are willing 143 00:05:31.845 --> 00:05:33.845 - to share data, especially, you know, there there 144 00:05:33.845 --> 00:05:34.165 - is, 145 00:05:35.045 --> 00:05:37.519 - a community of operators that tries to, you 146 00:05:37.519 --> 00:05:38.019 - know, 147 00:05:38.479 --> 00:05:40.879 - make fees make things as open as possible 148 00:05:40.879 --> 00:05:42.959 - and try to contribute to the community. And 149 00:05:42.959 --> 00:05:46.079 - these operators, generally, they are willing to to 150 00:05:46.079 --> 00:05:48.180 - share data. So for instance, historically, 151 00:05:48.959 --> 00:05:51.120 - there are these two platforms that are extremely 152 00:05:51.120 --> 00:05:53.295 - well known. It's Ripe Reese and RodViews, and 153 00:05:53.295 --> 00:05:54.035 - they actually, 154 00:05:54.495 --> 00:05:57.375 - are connected to, many, many networks. So these 155 00:05:57.375 --> 00:06:00.194 - operators, they are willing to to share data. 156 00:06:00.495 --> 00:06:02.254 - So now the goal with us is to 157 00:06:02.254 --> 00:06:04.589 - try to motivate even more. Like, you know, 158 00:06:04.589 --> 00:06:07.649 - to try to motivate operators that maybe before 159 00:06:08.110 --> 00:06:10.750 - they were not really interested into sharing data 160 00:06:10.750 --> 00:06:13.169 - to explain them, to motivate them, 161 00:06:13.629 --> 00:06:15.629 - to share more data. And now we are 162 00:06:15.629 --> 00:06:17.745 - starting to, in fact, see more and more 163 00:06:17.745 --> 00:06:19.665 - people that that are motivated. So this is 164 00:06:19.665 --> 00:06:21.185 - a good sign, and we will continue to 165 00:06:21.185 --> 00:06:22.704 - do this. Yeah. Oh, yeah. That's a very 166 00:06:22.704 --> 00:06:24.464 - good sign. And and I would say to 167 00:06:24.464 --> 00:06:26.625 - operators who are well, let's let's back up. 168 00:06:26.625 --> 00:06:28.704 - What is your pitch? How do you pitch 169 00:06:28.704 --> 00:06:30.144 - it? Because I know how I would pitch 170 00:06:30.144 --> 00:06:31.845 - it, but how do you pitch it? 171 00:06:32.680 --> 00:06:35.319 - Well, first, we we really well, it's not 172 00:06:35.319 --> 00:06:36.680 - so easy. And I think the pitch, you 173 00:06:36.680 --> 00:06:38.199 - know, it's a it's an it's not it's 174 00:06:38.199 --> 00:06:40.439 - not something you can just do, like, now 175 00:06:40.439 --> 00:06:42.939 - it's a long term project, bgprods.io. 176 00:06:43.240 --> 00:06:44.779 - But, basically, what we, 177 00:06:45.240 --> 00:06:47.745 - try to do is to explain them to 178 00:06:47.904 --> 00:06:50.144 - sort of educate them even though most of 179 00:06:50.144 --> 00:06:51.365 - them are already extremely, 180 00:06:51.745 --> 00:06:54.345 - knowledgeable. But, you know, recently, I went to 181 00:06:54.345 --> 00:06:56.464 - a lot of conferences like in The US. 182 00:06:56.464 --> 00:06:58.404 - I went to Nanog, I went to Aplicot, 183 00:06:59.264 --> 00:07:01.904 - I went to MINOG in the in in 184 00:07:01.904 --> 00:07:02.644 - in Dubai. 185 00:07:03.410 --> 00:07:05.970 - And and in each of these conference, I, 186 00:07:06.290 --> 00:07:08.209 - I I present bgproads.aio, 187 00:07:08.209 --> 00:07:10.689 - but I also talk about risk road views 188 00:07:10.689 --> 00:07:12.689 - because this is not just about bgproads.aio. 189 00:07:12.689 --> 00:07:15.169 - It's more generally about sharing bgp data. And 190 00:07:15.169 --> 00:07:17.564 - I really explained them that if we don't 191 00:07:17.564 --> 00:07:19.805 - have this data, if you imagine that you 192 00:07:19.805 --> 00:07:22.204 - live in a world where right priests and 193 00:07:22.204 --> 00:07:24.144 - road views, they just do not exist. 194 00:07:24.685 --> 00:07:27.004 - I mean, the Internet routing ecosystem would be 195 00:07:27.004 --> 00:07:29.564 - wild. There would be hijacks everywhere, world leaks, 196 00:07:29.564 --> 00:07:31.529 - and we would just don't know about them. 197 00:07:31.610 --> 00:07:33.689 - And this is very critical. So I'm really 198 00:07:33.689 --> 00:07:34.189 - explaining, 199 00:07:34.810 --> 00:07:35.310 - them, 200 00:07:36.009 --> 00:07:38.810 - this. And, and, yeah, it seems that they 201 00:07:38.810 --> 00:07:39.709 - sort of agree 202 00:07:40.250 --> 00:07:42.649 - that more and more they come and share 203 00:07:42.649 --> 00:07:43.149 - data, 204 00:07:43.529 --> 00:07:45.290 - but still a lot of a a lot 205 00:07:45.290 --> 00:07:46.910 - of operators, they just 206 00:07:47.574 --> 00:07:49.334 - they they sort of agree, but they just 207 00:07:49.334 --> 00:07:51.254 - don't take the time to to do this, 208 00:07:51.254 --> 00:07:53.435 - you know, and I understand that. 209 00:07:54.134 --> 00:07:55.495 - So with bgprs.au, 210 00:07:55.495 --> 00:07:57.254 - what you're also trying to do is to 211 00:07:57.254 --> 00:07:59.995 - simplify as much as possible the process also. 212 00:08:00.779 --> 00:08:02.620 - So they don't need to send us email 213 00:08:02.620 --> 00:08:04.620 - or anything. It's all automated. So we really 214 00:08:04.620 --> 00:08:07.020 - try on one side to educate, to explain 215 00:08:07.020 --> 00:08:08.860 - them why it's important, and on the other 216 00:08:08.860 --> 00:08:11.279 - side also to simplify their life as possible. 217 00:08:11.740 --> 00:08:13.680 - So that's what we're trying to do here. 218 00:08:13.964 --> 00:08:16.365 - Okay. Interesting. Yeah. My my pitch would be 219 00:08:16.365 --> 00:08:17.665 - along the lines of 220 00:08:18.524 --> 00:08:20.764 - you're actually not hiding what you think you're 221 00:08:20.764 --> 00:08:21.264 - hiding. 222 00:08:22.365 --> 00:08:25.264 - BGP is not built the way that allows 223 00:08:25.324 --> 00:08:25.824 - you 224 00:08:26.430 --> 00:08:27.329 - to hide 225 00:08:27.949 --> 00:08:30.029 - the information you think you can hide. You 226 00:08:30.029 --> 00:08:31.569 - think you can hide your peering. 227 00:08:32.190 --> 00:08:33.409 - Really, you can't. 228 00:08:33.950 --> 00:08:37.149 - And and but just just the nature of 229 00:08:37.149 --> 00:08:39.570 - BGP, you can't really hide your peering, 230 00:08:40.254 --> 00:08:41.774 - in the way that people think you can. 231 00:08:41.774 --> 00:08:44.334 - You can hide backup peers, but you can't 232 00:08:44.334 --> 00:08:46.495 - hide. And you can hide certain kinds of 233 00:08:46.495 --> 00:08:46.995 - paths 234 00:08:47.375 --> 00:08:49.154 - with certain ways of doing policy. 235 00:08:49.774 --> 00:08:51.294 - But you still have to say that I 236 00:08:51.294 --> 00:08:52.894 - have this reachable, and you still have to 237 00:08:52.894 --> 00:08:54.274 - say how you reach it 238 00:08:54.620 --> 00:08:56.299 - in order to be able to forward track. 239 00:08:56.299 --> 00:08:58.059 - You just like that's just part of being 240 00:08:58.059 --> 00:08:59.899 - able to forward traffic. By the way, we 241 00:08:59.899 --> 00:09:01.659 - got into this huge discussion when we were 242 00:09:01.659 --> 00:09:02.480 - talking about 243 00:09:02.940 --> 00:09:05.759 - SBGP versus past state vector versus SOBGP, 244 00:09:06.459 --> 00:09:08.379 - and a lot of the people objected to 245 00:09:08.379 --> 00:09:09.360 - things like SOBGP 246 00:09:09.980 --> 00:09:12.195 - and past a vector because they said, I 247 00:09:12.195 --> 00:09:14.115 - don't want everybody to know who I'm paired 248 00:09:14.115 --> 00:09:14.615 - with. 249 00:09:15.315 --> 00:09:17.254 - And, like, I don't I don't even understand. 250 00:09:18.035 --> 00:09:19.254 - I I don't even understand 251 00:09:20.355 --> 00:09:22.754 - why what how what do you think you're 252 00:09:22.754 --> 00:09:23.254 - hiding? 253 00:09:23.990 --> 00:09:25.449 - Like, I don't I don't comprehend 254 00:09:25.909 --> 00:09:27.610 - what you think you're doing there. 255 00:09:28.230 --> 00:09:29.909 - Well, there are still a lot of things 256 00:09:29.909 --> 00:09:30.970 - that, remains, 257 00:09:31.669 --> 00:09:33.610 - hidden in on the Internet. Right? 258 00:09:34.470 --> 00:09:34.970 - Actually, 259 00:09:35.350 --> 00:09:38.230 - yes. We do know, you know, some peering 260 00:09:38.230 --> 00:09:40.355 - links and and some of the information in 261 00:09:40.355 --> 00:09:43.154 - fact is available by default with BGP because 262 00:09:43.154 --> 00:09:45.574 - this is how BGP works. But today, 263 00:09:45.875 --> 00:09:48.855 - we have this information because we have ripries 264 00:09:48.995 --> 00:09:51.394 - and route views. Because if we don't have 265 00:09:51.394 --> 00:09:52.054 - this platform, 266 00:09:52.629 --> 00:09:54.789 - it would be much much harder to actually 267 00:09:54.789 --> 00:09:56.870 - know who is pairing with who. Yes you 268 00:09:56.870 --> 00:09:58.870 - do have your router and you you see 269 00:09:58.870 --> 00:10:00.870 - the ice pass but this is only a 270 00:10:00.870 --> 00:10:03.929 - very tiny fraction of what is actually happening 271 00:10:03.990 --> 00:10:04.809 - on the internet. 272 00:10:05.355 --> 00:10:07.195 - So if you don't have platform like Chris 273 00:10:07.195 --> 00:10:08.735 - Frodo with bgprose.io, 274 00:10:09.595 --> 00:10:11.355 - I think actually you would miss a lot 275 00:10:11.355 --> 00:10:12.475 - of information. So, 276 00:10:13.355 --> 00:10:14.074 - yeah. That's 277 00:10:16.475 --> 00:10:18.779 - And, of course, we are this is what 278 00:10:18.779 --> 00:10:20.860 - we show in the paper, that we published 279 00:10:20.860 --> 00:10:23.340 - at CECOM is that even with the high 280 00:10:23.340 --> 00:10:25.100 - number of vantage points that we have today, 281 00:10:25.100 --> 00:10:27.179 - we've reached reviews in all the different, 282 00:10:28.539 --> 00:10:31.360 - different platform like, of course, also PCH, 283 00:10:31.855 --> 00:10:34.575 - we are still already missing a lot of 284 00:10:34.575 --> 00:10:37.934 - very valuable information especially on peer to peer 285 00:10:37.934 --> 00:10:38.434 - links. 286 00:10:39.215 --> 00:10:42.014 - And this is why we want to to 287 00:10:42.014 --> 00:10:43.855 - build this also why why we want to 288 00:10:43.855 --> 00:10:44.894 - build this platform in 289 00:10:45.830 --> 00:10:48.549 - So what what, more vantage points. Sorry. Go 290 00:10:48.549 --> 00:10:50.549 - ahead. No. No. On the subject of vantage 291 00:10:50.549 --> 00:10:52.250 - points, what what kinds of operators 292 00:10:52.629 --> 00:10:55.370 - have been most enthusiastic? Are they eyeball networks? 293 00:10:55.429 --> 00:10:56.090 - Are they 294 00:10:56.470 --> 00:10:58.549 - content providers? Like, who who is it that's 295 00:10:58.549 --> 00:11:00.105 - most enthusiastic right now? 296 00:11:00.584 --> 00:11:02.264 - I mean, I think we have quite a 297 00:11:02.264 --> 00:11:03.964 - wide range of different, 298 00:11:04.504 --> 00:11:06.764 - operators. For instance, we have, 299 00:11:07.225 --> 00:11:07.725 - CDNs. 300 00:11:08.584 --> 00:11:10.845 - We are working with, OVH, 301 00:11:11.625 --> 00:11:14.524 - and they are pretty interested in, the project. 302 00:11:14.789 --> 00:11:15.990 - We also have, like, 303 00:11:16.950 --> 00:11:18.490 - people operating IXPs, 304 00:11:19.830 --> 00:11:20.570 - people operating, 305 00:11:20.950 --> 00:11:22.089 - let's say, regular 306 00:11:22.389 --> 00:11:23.370 - transit networks. 307 00:11:25.029 --> 00:11:25.190 - What 308 00:11:25.909 --> 00:11:27.394 - of course, right now, we do not have, 309 00:11:27.394 --> 00:11:30.674 - like, a direct peering from, any tier one 310 00:11:30.674 --> 00:11:30.995 - or, 311 00:11:31.875 --> 00:11:35.075 - but we still have peering from pretty large 312 00:11:35.075 --> 00:11:36.695 - networks, let's say, like, with 313 00:11:37.475 --> 00:11:39.815 - 100 of, of different peers. 314 00:11:40.924 --> 00:11:43.289 - And right now, like, just a few minutes 315 00:11:43.289 --> 00:11:43.789 - ago, 316 00:11:44.169 --> 00:11:45.049 - a guy from, 317 00:11:46.649 --> 00:11:48.909 - from South Africa just connected, 318 00:11:49.690 --> 00:11:52.730 - from an IXP, if I remember correctly. So 319 00:11:52.730 --> 00:11:54.375 - we have, like, quite a wide range of 320 00:11:54.615 --> 00:11:57.514 - different, network operators connecting to to the platform. 321 00:11:58.054 --> 00:11:59.095 - Yes. Although a lot of, 322 00:11:59.735 --> 00:12:02.774 - educational and hobby networks Yep. Like to connect 323 00:12:02.774 --> 00:12:03.914 - with us. That's sort 324 00:12:04.294 --> 00:12:05.754 - of obvious. But, yeah. 325 00:12:06.134 --> 00:12:08.455 - But but they are still providing some very 326 00:12:08.455 --> 00:12:09.899 - interesting information because, 327 00:12:10.379 --> 00:12:13.259 - we can get some very valuable information about 328 00:12:13.259 --> 00:12:14.080 - the peering, 329 00:12:14.700 --> 00:12:15.200 - links, 330 00:12:15.660 --> 00:12:18.620 - the, let's say, the local peering links that 331 00:12:18.620 --> 00:12:22.080 - are close to them. So we it's always 332 00:12:22.460 --> 00:12:23.759 - super interesting having 333 00:12:24.554 --> 00:12:26.715 - new peers. Yes. That's also by the way 334 00:12:26.715 --> 00:12:29.355 - just one difference between recent road views and 335 00:12:29.355 --> 00:12:31.995 - bgpros.io is that we really welcome everyone to 336 00:12:31.995 --> 00:12:34.475 - peer with us including these HOBY networks because 337 00:12:34.475 --> 00:12:36.634 - we think they're actually interesting point of view 338 00:12:36.634 --> 00:12:37.134 - sometimes. 339 00:12:37.940 --> 00:12:38.839 - So we really, 340 00:12:39.539 --> 00:12:41.539 - welcome everyone to appear with us. Yeah. I 341 00:12:41.539 --> 00:12:43.799 - was gonna say hobby networks and even 342 00:12:44.500 --> 00:12:46.519 - people who are at the edge traditional 343 00:12:47.700 --> 00:12:49.955 - enterprise, I don't I still don't like that 344 00:12:49.955 --> 00:12:51.235 - word, and I don't know what other word 345 00:12:51.235 --> 00:12:52.455 - to use. Mhmm. 346 00:12:53.475 --> 00:12:53.975 - Operators 347 00:12:54.595 --> 00:12:55.095 - are 348 00:12:56.595 --> 00:12:57.415 - still have 349 00:12:57.955 --> 00:13:01.075 - a lot of information that they're receiving from 350 00:13:01.075 --> 00:13:01.575 - upstream. 351 00:13:02.490 --> 00:13:04.649 - And they're particularly if they're dual home, particularly 352 00:13:04.649 --> 00:13:07.210 - if they're dual connected or quad connected, which 353 00:13:07.210 --> 00:13:09.610 - some companies are now quad connected, three and 354 00:13:09.610 --> 00:13:12.169 - four connections to the Internet instead of just 355 00:13:12.169 --> 00:13:12.669 - one. 356 00:13:13.049 --> 00:13:15.790 - And even if they're just using those connections 357 00:13:16.169 --> 00:13:16.669 - for 358 00:13:17.355 --> 00:13:19.754 - l three VPN or overlay of some type, 359 00:13:19.754 --> 00:13:22.315 - some type of SD WAN, you're still getting 360 00:13:22.475 --> 00:13:23.834 - a lot of times, you can still get 361 00:13:23.834 --> 00:13:24.654 - full routes, 362 00:13:25.034 --> 00:13:27.294 - and those full routes still give you perspectives 363 00:13:28.235 --> 00:13:30.574 - of where things are and stuff like that. 364 00:13:30.980 --> 00:13:32.199 - There are CDNs 365 00:13:33.059 --> 00:13:33.799 - who actually 366 00:13:34.339 --> 00:13:35.799 - hand customers boxes, 367 00:13:37.059 --> 00:13:39.620 - and the box is basically there to suck 368 00:13:39.620 --> 00:13:41.639 - in all the other peering 369 00:13:42.100 --> 00:13:42.600 - information 370 00:13:43.220 --> 00:13:46.035 - and ship that information back to the CDN 371 00:13:46.174 --> 00:13:46.674 - provider 372 00:13:47.215 --> 00:13:49.054 - to help them build a larger view of 373 00:13:49.054 --> 00:13:51.934 - the Internet. So that that data is, if 374 00:13:52.174 --> 00:13:53.315 - actually, very valuable 375 00:13:53.774 --> 00:13:54.754 - in many ways, 376 00:13:55.375 --> 00:13:56.595 - for that kind of stuff. 377 00:13:57.799 --> 00:13:59.959 - And you're just doing, yeah, you're just doing 378 00:13:59.959 --> 00:14:02.679 - BGP right now. Right? You're not doing, like, 379 00:14:02.679 --> 00:14:03.959 - like, Atlas where you're 380 00:14:04.679 --> 00:14:06.379 - we we are only doing, 381 00:14:06.919 --> 00:14:09.019 - control control plane measurements, 382 00:14:10.174 --> 00:14:12.355 - But we are not only doing BGP 383 00:14:13.054 --> 00:14:15.695 - in the way Riperes and Rodevus are doing 384 00:14:15.695 --> 00:14:16.674 - BGP because 385 00:14:17.134 --> 00:14:20.335 - our collector also supports BMP, which is the 386 00:14:20.335 --> 00:14:21.875 - BGP monitoring protocol. 387 00:14:22.735 --> 00:14:23.955 - Basically, that's just 388 00:14:24.539 --> 00:14:27.980 - TCP BGP data from all the peers connected 389 00:14:27.980 --> 00:14:30.459 - to the router through a TCP session. So 390 00:14:30.459 --> 00:14:33.339 - that gives us, like, a lot. Like, one 391 00:14:33.339 --> 00:14:35.019 - BMP session can give up 392 00:14:35.419 --> 00:14:37.054 - gives us, sorry, 393 00:14:37.595 --> 00:14:40.174 - a lot of very interesting information. 394 00:14:41.595 --> 00:14:43.995 - Oh, so you're so you're BMP. You're not 395 00:14:43.995 --> 00:14:45.294 - just peering BGP, 396 00:14:45.914 --> 00:14:48.634 - which is actually there's, yeah, there's a lot 397 00:14:48.634 --> 00:14:51.375 - more information in BMP Definitely. There is 398 00:14:51.679 --> 00:14:53.059 - in just a BGP peering. 399 00:14:53.759 --> 00:14:55.440 - For instance, a lot of people don't know 400 00:14:55.440 --> 00:14:57.199 - this, but one of the reasons the Internet 401 00:14:57.199 --> 00:14:59.779 - never converges, the DFC never converges, 402 00:15:00.480 --> 00:15:01.220 - is that 403 00:15:02.480 --> 00:15:02.980 - BGP 404 00:15:03.995 --> 00:15:04.495 - sends 405 00:15:05.034 --> 00:15:07.294 - does a lot more implicit withdraws 406 00:15:07.914 --> 00:15:10.235 - than people think when they first look at 407 00:15:10.235 --> 00:15:12.394 - it. They think Mhmm. Oh, you know, it 408 00:15:12.394 --> 00:15:14.315 - just sends an update every now and again. 409 00:15:14.315 --> 00:15:15.054 - No. Actually, 410 00:15:15.514 --> 00:15:17.855 - every time you change a community string 411 00:15:18.399 --> 00:15:19.220 - on a route 412 00:15:19.519 --> 00:15:21.220 - or you change the 413 00:15:21.519 --> 00:15:23.759 - the the peer that you're going through, even 414 00:15:23.759 --> 00:15:25.220 - if it's the same AS, 415 00:15:26.000 --> 00:15:29.279 - you you'll get implicit withdrawals because there's some 416 00:15:29.279 --> 00:15:30.659 - change about the path 417 00:15:31.095 --> 00:15:33.654 - that Exactly. Causes it. And you'll never see 418 00:15:33.654 --> 00:15:36.695 - that in a BGP feed. Right? You'll never 419 00:15:36.695 --> 00:15:39.014 - see that just in BGP. You can only 420 00:15:39.014 --> 00:15:41.414 - see that Mhmm. If you are one step 421 00:15:41.414 --> 00:15:42.714 - back or you're in BNB. 422 00:15:43.334 --> 00:15:45.434 - So those are, yeah, very interesting. 423 00:15:46.529 --> 00:15:49.409 - And, also, there are some events that do 424 00:15:49.409 --> 00:15:50.149 - not trigger 425 00:15:50.529 --> 00:15:53.570 - BGP updates even though they change the actual 426 00:15:53.570 --> 00:15:55.330 - state of the routing table, which is, like, 427 00:15:55.330 --> 00:15:55.909 - for instance, 428 00:15:56.450 --> 00:15:59.190 - a peering session that cuts at the TCP 429 00:15:59.250 --> 00:16:00.230 - level, for instance. 430 00:16:00.695 --> 00:16:02.475 - You will not necessarily get, 431 00:16:03.174 --> 00:16:04.794 - some new information 432 00:16:05.174 --> 00:16:06.315 - related to BGP, 433 00:16:07.414 --> 00:16:09.575 - even though it will actually change the state 434 00:16:09.575 --> 00:16:10.794 - of the routing table. 435 00:16:11.495 --> 00:16:12.475 - But with BMP, 436 00:16:13.090 --> 00:16:15.409 - this information is logged. So we have access 437 00:16:15.409 --> 00:16:17.490 - to this information. So this is also something 438 00:16:17.490 --> 00:16:19.909 - that is extremely valuable from BMP. 439 00:16:21.649 --> 00:16:23.809 - Yeah. Although I will say, I find it 440 00:16:23.809 --> 00:16:25.889 - even more fascinating that you're getting people to 441 00:16:25.889 --> 00:16:26.715 - pull give you BMP. 442 00:16:27.274 --> 00:16:30.415 - BNB because that's even that's even like, wow. 443 00:16:31.514 --> 00:16:34.254 - That's the management plane. That's like yeah. 444 00:16:34.634 --> 00:16:36.715 - There's there's actually yeah. There's a lot more 445 00:16:36.715 --> 00:16:38.495 - data there. So that's interesting. 446 00:16:40.190 --> 00:16:41.789 - And and by the way, for listeners who 447 00:16:41.789 --> 00:16:44.110 - don't know this, the reason these companies consider 448 00:16:44.110 --> 00:16:46.529 - this so important or such proprietary information 449 00:16:46.909 --> 00:16:48.509 - is that this is how they're doing their 450 00:16:48.509 --> 00:16:50.610 - traffic steering inbound and outbound 451 00:16:50.990 --> 00:16:52.370 - to do all of their 452 00:16:52.909 --> 00:16:53.409 - balancing 453 00:16:53.710 --> 00:16:54.210 - between 454 00:16:55.424 --> 00:16:58.804 - different link types. And, like, for a CDN, 455 00:16:59.585 --> 00:17:01.684 - this is how the CDN knows 456 00:17:02.465 --> 00:17:04.944 - where to ship traffic to make the path 457 00:17:04.944 --> 00:17:07.204 - as short as possible for optimal performance. 458 00:17:07.664 --> 00:17:09.345 - I shouldn't say short as possible. I should 459 00:17:09.345 --> 00:17:10.404 - say optimal performance. 460 00:17:10.865 --> 00:17:12.980 - So, So, yeah, that's why it's considered sensitive 461 00:17:12.980 --> 00:17:14.679 - a lot of times is because, like, 462 00:17:15.220 --> 00:17:17.779 - this is this is how their systems work 463 00:17:17.779 --> 00:17:20.660 - is basically doing this. So Yep. Sorry. That's 464 00:17:20.660 --> 00:17:21.619 - all. That's kind of like 465 00:17:22.339 --> 00:17:22.839 - okay. 466 00:17:23.380 --> 00:17:25.595 - So this is so the problem you're trying 467 00:17:25.595 --> 00:17:26.654 - to solve is 468 00:17:27.355 --> 00:17:28.095 - you're trying 469 00:17:28.674 --> 00:17:29.174 - to 470 00:17:29.754 --> 00:17:32.654 - get people to have or produce or create 471 00:17:33.434 --> 00:17:35.994 - a better view of the d f z, 472 00:17:35.994 --> 00:17:38.639 - of the default reason, of all the global 473 00:17:38.639 --> 00:17:40.259 - routing table as people say. 474 00:17:41.359 --> 00:17:42.899 - Okay. Yeah. Yeah. 475 00:17:43.759 --> 00:17:46.000 - But I think there is also, like, an 476 00:17:46.079 --> 00:17:48.880 - another objective of the bgparoutes.io 477 00:17:48.880 --> 00:17:51.734 - platform is that there are many different platforms 478 00:17:51.734 --> 00:17:54.794 - today like RACE, RODEWS, and they all have 479 00:17:55.095 --> 00:17:57.434 - different tools to access the data, different 480 00:17:57.974 --> 00:17:59.514 - dashboards, different API, 481 00:18:00.294 --> 00:18:01.115 - different stuffs. 482 00:18:01.654 --> 00:18:03.815 - And the goal of bgpruts.io 483 00:18:03.815 --> 00:18:04.875 - is also to 484 00:18:05.349 --> 00:18:05.849 - uniformize 485 00:18:06.549 --> 00:18:09.109 - the tool chain, that we have to access 486 00:18:09.109 --> 00:18:11.429 - the data. This is why bgpros.io 487 00:18:11.429 --> 00:18:13.509 - is actually mirroring the data from recent road 488 00:18:13.509 --> 00:18:14.009 - views 489 00:18:14.389 --> 00:18:16.569 - and write it in the same, 490 00:18:18.144 --> 00:18:19.125 - that in the same, 491 00:18:19.825 --> 00:18:21.605 - data format, sorry, 492 00:18:22.224 --> 00:18:24.785 - than the one collected from our BGP or 493 00:18:24.785 --> 00:18:26.325 - BMP sessions. So 494 00:18:27.105 --> 00:18:30.164 - therefore, it's easier for the users to access 495 00:18:30.549 --> 00:18:32.950 - different data, even though they actually come from 496 00:18:32.950 --> 00:18:33.450 - different, 497 00:18:34.470 --> 00:18:34.970 - platforms. 498 00:18:35.509 --> 00:18:38.170 - They only have one to chain to, 499 00:18:39.670 --> 00:18:41.450 - to use the data. That's also, 500 00:18:41.910 --> 00:18:43.609 - let's say secondary goal 501 00:18:44.164 --> 00:18:46.265 - to the to the platform. 502 00:18:46.964 --> 00:18:47.845 - Yes. What I'll, 503 00:18:48.325 --> 00:18:50.325 - what I like to say when I present 504 00:18:50.325 --> 00:18:50.724 - bgp 505 00:18:51.125 --> 00:18:52.164 - bgprs.io 506 00:18:52.164 --> 00:18:54.724 - is, like, it's good to collect more data, 507 00:18:54.724 --> 00:18:56.724 - but it's not so useful when we are 508 00:18:56.724 --> 00:18:59.039 - not equipped with the tools to process and 509 00:18:59.039 --> 00:19:01.919 - analyze all this data. So with bgprs.io 510 00:19:01.919 --> 00:19:04.000 - our goal is yes we want to collect 511 00:19:04.000 --> 00:19:06.319 - more data but we also want to provide 512 00:19:06.319 --> 00:19:07.539 - the tools to analyze 513 00:19:07.839 --> 00:19:08.659 - all this data 514 00:19:09.119 --> 00:19:10.980 - because in fact we actually found 515 00:19:11.615 --> 00:19:13.375 - when we did our research a couple of 516 00:19:13.375 --> 00:19:14.434 - years ago that 517 00:19:14.815 --> 00:19:16.755 - most of the operators and researchers 518 00:19:17.055 --> 00:19:19.855 - when they want to analyze BGP data, they 519 00:19:19.855 --> 00:19:22.414 - had to sample the data because they just 520 00:19:22.414 --> 00:19:24.174 - don't have the time or the resources to 521 00:19:24.174 --> 00:19:25.474 - process all this data 522 00:19:25.789 --> 00:19:28.109 - and that's typically what we want to avoid. 523 00:19:28.109 --> 00:19:29.869 - We want users to be able to just 524 00:19:29.869 --> 00:19:31.789 - process all this data, and this is why 525 00:19:31.789 --> 00:19:34.190 - we are also working on that side. That's 526 00:19:34.190 --> 00:19:36.269 - why we have the API and all these 527 00:19:36.269 --> 00:19:36.769 - dashboards. 528 00:19:38.109 --> 00:19:38.609 - Okay. 529 00:19:39.309 --> 00:19:39.809 - So 530 00:19:40.424 --> 00:19:40.924 - this 531 00:19:41.424 --> 00:19:41.924 - unified 532 00:19:42.424 --> 00:19:44.585 - data structure you're talking about, is it yang? 533 00:19:44.585 --> 00:19:46.345 - Is it something new? What have what have 534 00:19:46.345 --> 00:19:47.005 - you built? 535 00:19:48.664 --> 00:19:50.825 - So but yeah. You can go ahead. No. 536 00:19:50.825 --> 00:19:52.904 - Go ahead. Go ahead. Go ahead. Okay. Well, 537 00:19:52.904 --> 00:19:53.884 - we are basically, 538 00:19:54.539 --> 00:19:56.460 - getting all the data from risk road views 539 00:19:56.460 --> 00:19:58.860 - and our own collectors, and we are putting 540 00:19:58.860 --> 00:20:01.920 - everything into a database, like Postgres style database. 541 00:20:02.460 --> 00:20:03.680 - And then we have an API, 542 00:20:04.700 --> 00:20:06.779 - that is public. So you you can you 543 00:20:06.779 --> 00:20:08.755 - can query it if you want. And and 544 00:20:08.755 --> 00:20:10.355 - then there are a few endpoints if you 545 00:20:10.355 --> 00:20:12.194 - want you can get really the raw updates, 546 00:20:12.194 --> 00:20:14.674 - you can get rib entries at any point 547 00:20:14.674 --> 00:20:17.015 - in time, you can get also the topology 548 00:20:17.075 --> 00:20:18.855 - if you are interested into the topology, 549 00:20:19.795 --> 00:20:21.015 - and each endpoints 550 00:20:21.509 --> 00:20:23.849 - come with, comes with a set of parameters. 551 00:20:23.990 --> 00:20:26.009 - So you can really say, okay, I want 552 00:20:26.230 --> 00:20:28.650 - these updates from these vantage points 553 00:20:28.950 --> 00:20:29.930 - where IS, 554 00:20:30.390 --> 00:20:31.269 - 3333 555 00:20:31.269 --> 00:20:32.869 - appears in the IS pass. And you will 556 00:20:32.869 --> 00:20:34.410 - just get these updates. 557 00:20:35.204 --> 00:20:37.125 - And there are many other parameters like that, 558 00:20:37.125 --> 00:20:39.525 - so you can really breed flexible queries and 559 00:20:39.525 --> 00:20:41.684 - just get the data you need. So you 560 00:20:41.684 --> 00:20:44.005 - don't need to download, for instance, terabytes of 561 00:20:44.005 --> 00:20:46.325 - data if you want to analyze something, you 562 00:20:46.325 --> 00:20:48.325 - can just get the slice of the data 563 00:20:48.325 --> 00:20:50.220 - you need. That's what we are trying to 564 00:20:50.220 --> 00:20:51.599 - do and that simplifies 565 00:20:52.059 --> 00:20:54.720 - a lot the life of operators and researchers 566 00:20:55.099 --> 00:20:57.420 - and it makes all these analyses more accurate 567 00:20:57.420 --> 00:20:58.720 - and faster as well. 568 00:20:59.259 --> 00:21:01.500 - So how far how far back in time 569 00:21:01.500 --> 00:21:03.279 - do you aspire to go? 570 00:21:04.144 --> 00:21:05.444 - Yes. So Yep. 571 00:21:05.904 --> 00:21:08.325 - So here in our case, we 572 00:21:09.265 --> 00:21:09.765 - unlike 573 00:21:10.224 --> 00:21:11.125 - recent Rogers, 574 00:21:11.585 --> 00:21:12.644 - who has, like, 575 00:21:13.024 --> 00:21:16.804 - years, decay even years, decades of historical data, 576 00:21:17.184 --> 00:21:18.420 - here we aim at, 577 00:21:21.140 --> 00:21:22.200 - let's say, preferring 578 00:21:22.660 --> 00:21:23.160 - coverage 579 00:21:23.539 --> 00:21:24.920 - over data retention. 580 00:21:25.380 --> 00:21:27.960 - So for now, we are retaining for 581 00:21:28.580 --> 00:21:30.200 - four months of data. 582 00:21:31.144 --> 00:21:32.904 - Of course, we can we could do more 583 00:21:32.904 --> 00:21:34.984 - if we had more disk, but, 584 00:21:35.704 --> 00:21:38.024 - it takes it takes a lot of space, 585 00:21:38.024 --> 00:21:38.524 - obviously. 586 00:21:39.704 --> 00:21:42.605 - But also what we saw from our logs 587 00:21:42.664 --> 00:21:43.325 - is that 588 00:21:43.784 --> 00:21:45.884 - usually operators, they don't go 589 00:21:46.700 --> 00:21:48.400 - deeper than, like, one month 590 00:21:49.259 --> 00:21:52.220 - ago. So Uh-huh. I think storing data for, 591 00:21:52.220 --> 00:21:53.279 - like, a few months, 592 00:21:53.819 --> 00:21:55.500 - in most of the use cases, not only, 593 00:21:55.500 --> 00:21:57.119 - of course, if you want to do historical 594 00:21:57.180 --> 00:22:00.319 - analysis, like, for BGP hijacks, route leaks, whatever, 595 00:22:00.585 --> 00:22:02.044 - you will need, of course, 596 00:22:02.345 --> 00:22:03.085 - the historical 597 00:22:03.384 --> 00:22:05.644 - data from recent Rogers. And this is why 598 00:22:06.105 --> 00:22:08.044 - these platforms are extremely useful. 599 00:22:08.345 --> 00:22:10.265 - The goal of BGP Rosario is not to 600 00:22:10.265 --> 00:22:11.565 - replace recent Rogers 601 00:22:12.105 --> 00:22:13.644 - but to propose a complementary 602 00:22:14.025 --> 00:22:14.525 - approach 603 00:22:15.480 --> 00:22:15.980 - where 604 00:22:16.440 --> 00:22:17.339 - we favoritize 605 00:22:17.640 --> 00:22:19.319 - I was looking for the word favoritize 606 00:22:20.119 --> 00:22:20.859 - the coverage 607 00:22:21.319 --> 00:22:22.619 - over the data retention. 608 00:22:24.119 --> 00:22:26.599 - So here for now it's I think the 609 00:22:26.599 --> 00:22:28.619 - parameter is set at three months, 610 00:22:29.985 --> 00:22:31.325 - but if one day 611 00:22:31.705 --> 00:22:33.644 - it turns out that operators are interested, 612 00:22:34.825 --> 00:22:38.105 - in having more historical data, like six months, 613 00:22:38.105 --> 00:22:40.505 - one year, we will have to tune this, 614 00:22:40.664 --> 00:22:43.119 - this parameter. Of course. And are you are 615 00:22:43.119 --> 00:22:45.299 - you archiving it, like, on, 616 00:22:45.839 --> 00:22:47.920 - like, archival media or anything, or is it 617 00:22:47.920 --> 00:22:49.380 - just gone after four months? 618 00:22:51.440 --> 00:22:53.805 - For now, unfortunately, it's gone after four months. 619 00:22:54.045 --> 00:22:54.545 - Yeah. 620 00:22:55.005 --> 00:22:58.365 - But, we are also not only exposing the 621 00:22:58.365 --> 00:22:59.345 - data through 622 00:23:00.845 --> 00:23:01.585 - our API, 623 00:23:01.965 --> 00:23:04.865 - but also through, let's say, classic MRT 624 00:23:05.244 --> 00:23:05.744 - archives 625 00:23:06.349 --> 00:23:08.690 - that we published every five minutes 626 00:23:09.230 --> 00:23:09.970 - on our, 627 00:23:10.830 --> 00:23:11.330 - on 628 00:23:11.710 --> 00:23:12.930 - a website that we have 629 00:23:13.549 --> 00:23:15.250 - with the the MRT archives. 630 00:23:15.549 --> 00:23:18.210 - But perhaps we can think about, like, archiving 631 00:23:18.590 --> 00:23:21.390 - these MRT archives that are more than four 632 00:23:21.390 --> 00:23:21.890 - months 633 00:23:22.585 --> 00:23:23.085 - old 634 00:23:23.464 --> 00:23:25.484 - into other way, storage, 635 00:23:26.184 --> 00:23:26.825 - strategies. But, 636 00:23:27.704 --> 00:23:30.044 - to to be honest, we didn't thought about 637 00:23:30.264 --> 00:23:31.464 - did not think about this, 638 00:23:32.105 --> 00:23:33.484 - much, much more. 639 00:23:33.990 --> 00:23:35.750 - Well, it costs money to store bits. It's 640 00:23:35.750 --> 00:23:39.529 - not free. So Yeah. Exactly. Exactly. Unfortunately, we 641 00:23:39.990 --> 00:23:41.130 - do not have, like, 642 00:23:41.509 --> 00:23:42.730 - hundreds of terabytes, 643 00:23:43.509 --> 00:23:44.250 - of disks. 644 00:23:44.789 --> 00:23:46.710 - So, yeah, fortunately, we have to do some, 645 00:23:47.029 --> 00:23:47.849 - trade offs. 646 00:23:48.470 --> 00:23:48.970 - Yep. 647 00:23:50.095 --> 00:23:51.954 - So and you just store this 648 00:23:52.575 --> 00:23:55.054 - basically full data for three or four months, 649 00:23:55.054 --> 00:23:56.115 - and then it's gone. 650 00:23:56.734 --> 00:23:59.615 - Have you ever considered as well, just another 651 00:23:59.615 --> 00:24:01.075 - idea to throw out there, 652 00:24:04.480 --> 00:24:04.980 - Aggregating 653 00:24:05.440 --> 00:24:08.400 - or summarizing the data in some way after 654 00:24:08.400 --> 00:24:11.299 - the four months so that you have 10% 655 00:24:11.759 --> 00:24:12.660 - usage of 656 00:24:13.039 --> 00:24:15.359 - the same space or something like that. So 657 00:24:15.359 --> 00:24:15.884 - you're actually, 658 00:24:20.765 --> 00:24:21.265 - reducing 659 00:24:21.565 --> 00:24:23.724 - the amount of storage you're using while trying 660 00:24:23.724 --> 00:24:26.224 - to keep it maybe a bit more useful. 661 00:24:26.525 --> 00:24:27.964 - I don't know. That's just a matter. This 662 00:24:27.964 --> 00:24:30.450 - is definitely something we thought about because, actually, 663 00:24:30.450 --> 00:24:32.950 - the paper that we published at CECOM is 664 00:24:33.250 --> 00:24:34.549 - exactly this. 665 00:24:35.650 --> 00:24:37.349 - The purpose of this paper is 666 00:24:37.970 --> 00:24:40.470 - which BGP update should we keep 667 00:24:41.329 --> 00:24:41.829 - to 668 00:24:42.865 --> 00:24:44.644 - keep the useful 669 00:24:45.025 --> 00:24:47.845 - information while reducing the volume of the data. 670 00:24:48.464 --> 00:24:51.125 - So what, I mean, this is not implemented 671 00:24:51.184 --> 00:24:53.184 - yet, but this is definitely something that we 672 00:24:53.184 --> 00:24:53.845 - can implement, 673 00:24:54.464 --> 00:24:57.789 - is applying this reduction, this lossy 674 00:24:58.490 --> 00:24:59.549 - compression algorithm 675 00:25:01.369 --> 00:25:02.269 - to the 676 00:25:02.730 --> 00:25:05.150 - data that is more than four 677 00:25:05.529 --> 00:25:06.509 - months old. 678 00:25:07.049 --> 00:25:07.950 - Definitely. But 679 00:25:08.265 --> 00:25:11.224 - yet, unfortunately, we have, like, still a lot 680 00:25:11.224 --> 00:25:12.205 - of things to 681 00:25:12.664 --> 00:25:14.505 - to do to improve the platform. So this 682 00:25:14.505 --> 00:25:16.664 - is not implemented yet, but this is definitely 683 00:25:16.664 --> 00:25:17.724 - something that we 684 00:25:18.025 --> 00:25:20.285 - thought about. And also another paper 685 00:25:20.585 --> 00:25:21.565 - that we published, 686 00:25:21.960 --> 00:25:23.640 - I mean, it's not a paper, it's a 687 00:25:23.640 --> 00:25:25.019 - poster at IMC, 688 00:25:26.039 --> 00:25:26.779 - is about 689 00:25:27.400 --> 00:25:27.900 - selecting 690 00:25:29.159 --> 00:25:31.179 - the vantage points that are the most valuable 691 00:25:31.240 --> 00:25:32.460 - and by valuable, 692 00:25:33.000 --> 00:25:34.220 - that are the most 693 00:25:35.045 --> 00:25:35.924 - dissimilar. Sorry. 694 00:25:38.164 --> 00:25:40.505 - And this is perhaps also something that, 695 00:25:41.204 --> 00:25:44.085 - we can use to keep only data from 696 00:25:44.085 --> 00:25:45.045 - 5% 697 00:25:45.045 --> 00:25:46.484 - of the vantage points that are the most 698 00:25:46.484 --> 00:25:46.984 - valuable, 699 00:25:47.525 --> 00:25:50.005 - for data that is more than four months 700 00:25:50.005 --> 00:25:50.430 - old. 701 00:25:51.549 --> 00:25:53.869 - But, but yeah. Well, once again, this is 702 00:25:53.869 --> 00:25:55.090 - not implemented yet. 703 00:25:56.830 --> 00:25:58.930 - Do you find that the people consuming 704 00:25:59.549 --> 00:26:01.490 - the service, do they have a lot of 705 00:26:01.789 --> 00:26:03.549 - requests or are they just like, oh, yeah. 706 00:26:03.549 --> 00:26:05.070 - You've got, like, three or four endpoints that 707 00:26:05.070 --> 00:26:06.535 - I use all the time. That's what I 708 00:26:06.535 --> 00:26:08.695 - go after. Do you have people asking for 709 00:26:08.695 --> 00:26:09.434 - different views? 710 00:26:09.974 --> 00:26:10.875 - Stuff like that. 711 00:26:11.494 --> 00:26:13.335 - We we do have, people that, 712 00:26:14.375 --> 00:26:16.375 - queries, like, all the Vonage points. 713 00:26:16.775 --> 00:26:19.275 - Sometimes it's tenants. Sometimes it's operators. 714 00:26:20.380 --> 00:26:22.640 - Generally, they're interested like by 715 00:26:23.019 --> 00:26:25.660 - real time or short term historical data. They 716 00:26:25.660 --> 00:26:26.480 - don't go like 717 00:26:27.179 --> 00:26:28.559 - like three four years 718 00:26:28.940 --> 00:26:29.440 - back. 719 00:26:29.819 --> 00:26:32.059 - For some reason, this is appears to be 720 00:26:32.059 --> 00:26:33.920 - not so useful for them. So okay. 721 00:26:34.335 --> 00:26:37.054 - That's also why we decided not to devote 722 00:26:37.054 --> 00:26:37.954 - time on that. 723 00:26:39.054 --> 00:26:40.654 - But yeah. Yeah. Generally, they they they they 724 00:26:40.654 --> 00:26:42.674 - are interested in you to that. They also, 725 00:26:43.134 --> 00:26:44.434 - just go to our dashboards. 726 00:26:44.734 --> 00:26:45.554 - So for instance, 727 00:26:46.920 --> 00:26:49.000 - they don't necessarily need to use the API 728 00:26:49.000 --> 00:26:50.840 - and to really analyze the raw data. Now 729 00:26:50.840 --> 00:26:52.299 - we have a set of dashboards 730 00:26:52.920 --> 00:26:53.320 - which, 731 00:26:53.799 --> 00:26:54.920 - displays a lot of, 732 00:26:55.559 --> 00:26:56.059 - like, 733 00:26:56.680 --> 00:26:59.880 - metrics about, which prefix is visible from which 734 00:26:59.880 --> 00:27:00.779 - Vonage points, 735 00:27:01.625 --> 00:27:02.744 - which peering links, 736 00:27:03.144 --> 00:27:05.544 - exist between these ICS and so on and 737 00:27:05.544 --> 00:27:07.784 - so forth. So, generally, they don't just go 738 00:27:07.784 --> 00:27:08.684 - to these dashboards, 739 00:27:09.544 --> 00:27:11.464 - and and and that's, that's what we see. 740 00:27:11.464 --> 00:27:12.684 - So we see many more, 741 00:27:13.384 --> 00:27:14.924 - users going to these dashboards 742 00:27:15.329 --> 00:27:17.409 - compared to users using our API, but still, 743 00:27:17.409 --> 00:27:19.169 - there are some really people that are really 744 00:27:19.169 --> 00:27:19.669 - motivated, 745 00:27:20.529 --> 00:27:22.390 - that actually use the API. Yes. 746 00:27:24.609 --> 00:27:26.069 - And our most, 747 00:27:26.450 --> 00:27:27.589 - our most famous, 748 00:27:27.890 --> 00:27:31.315 - successful, let's say, dashboard is the AS explorer 749 00:27:31.315 --> 00:27:34.695 - dashboard that displays various information about 750 00:27:34.995 --> 00:27:36.535 - a specific AS, like, 751 00:27:36.995 --> 00:27:40.535 - what are its relationships with the other ASs, 752 00:27:40.674 --> 00:27:44.134 - what prefix is prefixes is it announcing, 753 00:27:45.230 --> 00:27:47.549 - what is in the who is database for 754 00:27:47.549 --> 00:27:48.529 - this ASN. 755 00:27:49.070 --> 00:27:50.509 - Well, several other, 756 00:27:51.230 --> 00:27:53.070 - other information. And this one is the most, 757 00:27:53.309 --> 00:27:54.850 - the most popular yet. 758 00:27:55.869 --> 00:27:57.549 - I could see this as an operator, I 759 00:27:57.549 --> 00:27:59.710 - could see this being super useful for just 760 00:27:59.710 --> 00:28:01.525 - an audit. Like, what what do I look 761 00:28:01.525 --> 00:28:04.244 - like to everybody else? Because you've already gone 762 00:28:04.244 --> 00:28:05.525 - and collected it all. It's not like I 763 00:28:05.525 --> 00:28:06.884 - have to go log in to route servers 764 00:28:06.884 --> 00:28:08.904 - and looking glasses and all kinds of stuff. 765 00:28:09.125 --> 00:28:12.025 - Yeah. This actually helped, some operators 766 00:28:12.404 --> 00:28:14.505 - already to find a route leak 767 00:28:14.869 --> 00:28:15.369 - because, 768 00:28:15.910 --> 00:28:18.250 - we are inferring the AS relationships 769 00:28:18.549 --> 00:28:19.350 - like with, 770 00:28:20.150 --> 00:28:21.130 - of course, a probabilistic 771 00:28:21.590 --> 00:28:22.090 - algorithm. 772 00:28:22.789 --> 00:28:25.529 - And one AS appeared as upstream 773 00:28:25.830 --> 00:28:26.390 - of the, 774 00:28:27.029 --> 00:28:29.350 - the AS of the operator checking the dashboard. 775 00:28:29.350 --> 00:28:30.484 - He said, oh, no. No. This guy is 776 00:28:30.484 --> 00:28:32.244 - not supposed to be my upstream. It's supposed 777 00:28:32.244 --> 00:28:33.144 - to be my downstream. 778 00:28:33.605 --> 00:28:36.404 - And then we show the AS path that 779 00:28:36.404 --> 00:28:38.244 - led to this inference. And he said, oh, 780 00:28:38.244 --> 00:28:39.764 - no. No. This is not supposed to happen. 781 00:28:39.764 --> 00:28:42.005 - This is a route leak. And so, actually, 782 00:28:42.005 --> 00:28:44.589 - it helped the operators to to found an 783 00:28:44.589 --> 00:28:48.509 - actual, BGP anomaly. So yeah. Yeah. Cool. Yeah. 784 00:28:48.509 --> 00:28:50.109 - So I was gonna ask, like, what are 785 00:28:50.109 --> 00:28:51.470 - some of the tools that you have? So 786 00:28:51.470 --> 00:28:54.609 - that's clearly one of them. Right? Is this 787 00:28:54.669 --> 00:28:55.809 - is this dashboard 788 00:28:56.384 --> 00:28:59.105 - are there other tools that people use off 789 00:28:59.105 --> 00:29:00.644 - of this? Do you have, like, 790 00:29:00.945 --> 00:29:03.345 - a store? Can you go to, like, the 791 00:29:03.345 --> 00:29:05.105 - bgproutes.io 792 00:29:05.105 --> 00:29:06.404 - store and download? 793 00:29:08.785 --> 00:29:11.309 - I'm I'm laughing because this is exactly the 794 00:29:11.309 --> 00:29:13.569 - idea we had when we built the platform. 795 00:29:14.509 --> 00:29:16.829 - But I I will maybe let Thomas, he's 796 00:29:16.829 --> 00:29:18.289 - more an expert in the dashboards, 797 00:29:18.589 --> 00:29:20.529 - in the logs of the dashboards. So 798 00:29:21.710 --> 00:29:23.390 - Yeah. Well, we have, as I was saying, 799 00:29:23.390 --> 00:29:25.684 - we have a set of, different dashboards. We 800 00:29:25.684 --> 00:29:27.765 - have the AI explorer that Thomas just described. 801 00:29:27.765 --> 00:29:29.204 - We have also a looking glass. You know, 802 00:29:29.204 --> 00:29:31.684 - looking glass, it's a very common dashboard. There 803 00:29:31.684 --> 00:29:33.785 - is also the you the the Eureka electric 804 00:29:33.845 --> 00:29:36.164 - looking glass that perhaps you know. There are 805 00:29:36.164 --> 00:29:38.664 - a few other looking glasses available, online. 806 00:29:39.759 --> 00:29:42.000 - So, yes, it's yet another looking glass that 807 00:29:42.000 --> 00:29:44.079 - we provide, but this time, really, we want 808 00:29:44.079 --> 00:29:46.640 - to show the routes from all the vantage 809 00:29:46.640 --> 00:29:49.059 - points from reach route, give us bgproutes.io. 810 00:29:49.359 --> 00:29:50.799 - So this is really the goal we have 811 00:29:50.799 --> 00:29:52.259 - is to have this high coverage 812 00:29:52.795 --> 00:29:54.954 - and to give as much information as possible. 813 00:29:54.954 --> 00:29:57.134 - So we also, for instance, show the, 814 00:29:57.434 --> 00:30:00.255 - root origin validation status. We show also 815 00:30:00.634 --> 00:30:01.775 - in the ice explorer, 816 00:30:02.474 --> 00:30:04.315 - and also in the looking glass, we show 817 00:30:04.315 --> 00:30:06.795 - the ASPA validation status. This is something really 818 00:30:06.795 --> 00:30:09.230 - new. ASPA, it's it's been there for just 819 00:30:09.230 --> 00:30:11.230 - a few months, a few years. So we're 820 00:30:11.230 --> 00:30:12.909 - also doing this. So we really try to 821 00:30:12.909 --> 00:30:15.329 - integrate the latest technology that exists, 822 00:30:16.109 --> 00:30:18.269 - into our dashboards and our, 823 00:30:18.990 --> 00:30:19.970 - monitoring tools. 824 00:30:21.575 --> 00:30:23.734 - We also have IS rank. So, you know, 825 00:30:23.734 --> 00:30:25.195 - operators, they like to, 826 00:30:25.975 --> 00:30:28.394 - compare their network with the other networks. 827 00:30:28.855 --> 00:30:30.795 - So we have also IS rank. 828 00:30:31.174 --> 00:30:33.734 - It's not the most popular dashboard, definitely not, 829 00:30:33.734 --> 00:30:36.230 - but we see still many operators that that 830 00:30:36.230 --> 00:30:36.730 - regularly 831 00:30:37.109 --> 00:30:39.210 - check their ranking, so that's that's nice. 832 00:30:39.750 --> 00:30:40.250 - And, 833 00:30:41.670 --> 00:30:44.390 - and soon we will also release, probably next 834 00:30:44.390 --> 00:30:46.869 - week, a new dashboard which is called the 835 00:30:46.869 --> 00:30:47.369 - IRR 836 00:30:47.670 --> 00:30:50.070 - Explorer. So IRR, you know, it's the Internet 837 00:30:50.070 --> 00:30:50.890 - Lending Registries. 838 00:30:51.585 --> 00:30:54.724 - It's something that is used by operators to, 839 00:30:55.424 --> 00:30:58.304 - explain their policies, you know, in in some 840 00:30:58.304 --> 00:30:59.765 - objects and also to, 841 00:31:00.464 --> 00:31:02.625 - apply some filters in their routers. So this 842 00:31:02.625 --> 00:31:04.625 - is very useful data, and now it's a 843 00:31:04.625 --> 00:31:06.865 - bit complicated to navigate through all these data. 844 00:31:06.865 --> 00:31:08.519 - So we are bringing a dashboard that will 845 00:31:08.519 --> 00:31:10.440 - help users to to navigate through all this 846 00:31:10.440 --> 00:31:13.019 - data. So the goal with these dashboards is, 847 00:31:13.480 --> 00:31:15.019 - one, to help operators 848 00:31:15.399 --> 00:31:19.000 - to, you know, monitor BGP, but it's also 849 00:31:19.000 --> 00:31:21.880 - to show them why it matters to share 850 00:31:21.880 --> 00:31:23.099 - data. Because 851 00:31:23.664 --> 00:31:26.384 - all these dashboards, they are all possible because 852 00:31:26.384 --> 00:31:28.625 - we have this data. And I think, you 853 00:31:28.625 --> 00:31:29.125 - know, 854 00:31:29.664 --> 00:31:31.744 - if we don't have these dashboards, operators, they 855 00:31:31.744 --> 00:31:33.345 - just connect to us and they they don't 856 00:31:33.345 --> 00:31:35.025 - see anything. They they raise it there is 857 00:31:35.025 --> 00:31:37.259 - no motivation to come and add more data. 858 00:31:37.259 --> 00:31:39.019 - But if we show them that it's useful 859 00:31:39.019 --> 00:31:41.180 - to do it, then maybe, you know, it 860 00:31:41.180 --> 00:31:42.859 - will be like a snowball, and and and 861 00:31:42.859 --> 00:31:44.539 - we will get more and more people because 862 00:31:44.539 --> 00:31:46.320 - they like it. You know? So yeah. 863 00:31:47.099 --> 00:31:47.920 - May maybe, 864 00:31:48.380 --> 00:31:49.359 - perhaps, like, 865 00:31:49.820 --> 00:31:50.730 - a a little 866 00:31:51.744 --> 00:31:53.605 - thing about our looking glass because 867 00:31:54.464 --> 00:31:56.144 - I I think that our looking glass is 868 00:31:56.144 --> 00:31:58.085 - not yet another looking glass, 869 00:31:58.544 --> 00:32:01.924 - because usually looking glass shows all the routes, 870 00:32:03.505 --> 00:32:05.524 - that are installed in one router. 871 00:32:06.150 --> 00:32:08.789 - Here, the idea of our Looking Glass is 872 00:32:08.789 --> 00:32:11.750 - to see one prefix, but from all the 873 00:32:11.750 --> 00:32:14.869 - different vantage points, to see how the different 874 00:32:14.869 --> 00:32:16.570 - people in the Internet reaches 875 00:32:17.190 --> 00:32:18.744 - reach, sorry, this 876 00:32:19.125 --> 00:32:19.625 - specific 877 00:32:19.924 --> 00:32:22.025 - prefix. So it can also help, 878 00:32:22.725 --> 00:32:26.025 - the operators to see how far their prefixes 879 00:32:26.164 --> 00:32:26.985 - are propagating. 880 00:32:28.005 --> 00:32:29.940 - Maybe if this prefix is seen by, like, 881 00:32:30.179 --> 00:32:31.059 - 99% 882 00:32:31.059 --> 00:32:32.899 - of the vantage points, well, it's fine. But 883 00:32:32.899 --> 00:32:34.919 - if it's seen by, like, 70%, 884 00:32:35.460 --> 00:32:37.879 - perhaps there there is something wrong. So 885 00:32:38.500 --> 00:32:40.359 - this is the the goal of this dashboard. 886 00:32:41.139 --> 00:32:42.279 - It's, like, 887 00:32:42.694 --> 00:32:44.694 - this is why we call it, like, the 888 00:32:44.694 --> 00:32:48.315 - prefix explorer, not really the the looking glass. 889 00:32:48.855 --> 00:32:49.914 - It's a bit different. 890 00:32:51.575 --> 00:32:53.894 - So is there, oh, sorry. Go ahead, Erase. 891 00:32:53.894 --> 00:32:55.890 - No. Go ahead, Tom. Go. Oh, I I 892 00:32:55.890 --> 00:32:57.910 - was just gonna ask. So there's so there's 893 00:32:57.970 --> 00:32:59.970 - already a rich set of BGP data out 894 00:32:59.970 --> 00:33:01.730 - there. Right? We have IRRs. We have Looking 895 00:33:01.730 --> 00:33:02.789 - Glass. We have, 896 00:33:03.490 --> 00:33:06.049 - risk and route views. So have you have 897 00:33:06.049 --> 00:33:08.369 - you noticed anything? Because it sounds like what 898 00:33:08.369 --> 00:33:10.994 - you're starting to do is correlate all these 899 00:33:10.994 --> 00:33:11.815 - things together, 900 00:33:12.434 --> 00:33:14.035 - and, you know, maybe it's not all fully 901 00:33:14.035 --> 00:33:16.115 - integrated yet. But if you had if you 902 00:33:16.115 --> 00:33:18.835 - had a policy view from from an IRR 903 00:33:18.835 --> 00:33:21.075 - and if you had a looking glass and 904 00:33:21.075 --> 00:33:22.960 - this other looking glass and, like, if you 905 00:33:22.960 --> 00:33:24.400 - if you have all that data together, it 906 00:33:24.400 --> 00:33:25.840 - seems like you might be able to draw 907 00:33:25.840 --> 00:33:26.580 - some conclusions 908 00:33:26.960 --> 00:33:29.039 - that were not possible with the data isolated. 909 00:33:29.039 --> 00:33:31.519 - Right? Any any examples of that or thoughts 910 00:33:31.519 --> 00:33:32.259 - on that? 911 00:33:33.039 --> 00:33:34.960 - Yeah. Well, I can give you an example. 912 00:33:34.960 --> 00:33:37.884 - So for instance, from the IRR, you have 913 00:33:37.884 --> 00:33:40.444 - the Outnume object and from the content in 914 00:33:40.444 --> 00:33:42.545 - the Outnume object you can infer 915 00:33:43.805 --> 00:33:44.625 - ICE relationships 916 00:33:45.005 --> 00:33:46.944 - and you can infer if a relationship 917 00:33:47.244 --> 00:33:47.744 - is 918 00:33:48.045 --> 00:33:50.065 - a customer, a provider or a peer 919 00:33:51.929 --> 00:33:53.149 - Of course, it's probabilistic, 920 00:33:53.450 --> 00:33:55.149 - but, with a pretty decent, 921 00:33:56.089 --> 00:33:58.329 - like, accuracy, you can do that. And then 922 00:33:58.329 --> 00:33:59.769 - you can do the same. You can do 923 00:33:59.769 --> 00:34:02.429 - IS relationship inference using the BGP data, 924 00:34:02.970 --> 00:34:04.835 - and then you can correlate things. So For 925 00:34:04.835 --> 00:34:07.075 - instance, if in BGP you see that this 926 00:34:07.075 --> 00:34:08.914 - guy is supposed to be your customer, and 927 00:34:08.914 --> 00:34:10.675 - if you look at the IRR, you see, 928 00:34:10.675 --> 00:34:12.275 - oh, no, it's supposed to be a provider, 929 00:34:12.275 --> 00:34:14.034 - and so there is a mismatch, then maybe 930 00:34:14.034 --> 00:34:14.934 - there is a misconfiguration 931 00:34:15.315 --> 00:34:17.235 - in the IRR, or maybe there is something 932 00:34:17.235 --> 00:34:17.974 - wrong in BGP. 933 00:34:18.410 --> 00:34:20.250 - And now additionally on top of that, you 934 00:34:20.250 --> 00:34:21.550 - can even look at ASFA. 935 00:34:21.849 --> 00:34:23.210 - You can look at ASFA. You can say, 936 00:34:23.210 --> 00:34:25.210 - okay, what does ASFA says? And then, 937 00:34:25.849 --> 00:34:27.849 - maybe ASFA will confirm what you see in 938 00:34:27.849 --> 00:34:29.610 - BGP or in the IR, and then you 939 00:34:29.610 --> 00:34:32.085 - can really debug. So what you said is 940 00:34:32.085 --> 00:34:33.204 - is 100% 941 00:34:33.204 --> 00:34:35.284 - true. The goal is really to correlate all 942 00:34:35.284 --> 00:34:36.025 - this information 943 00:34:36.485 --> 00:34:38.485 - to try to help operators to find, for 944 00:34:38.485 --> 00:34:39.465 - instance, misconfigurations, 945 00:34:40.644 --> 00:34:41.144 - problems, 946 00:34:41.525 --> 00:34:43.364 - lack of connectivity, and so on and so 947 00:34:43.364 --> 00:34:44.965 - forth. And in fact, coding all all this 948 00:34:44.965 --> 00:34:46.264 - data is definitely useful. 949 00:34:46.889 --> 00:34:49.210 - Yeah. And also right now we are working 950 00:34:49.210 --> 00:34:49.710 - on 951 00:34:50.409 --> 00:34:54.250 - the anomaly detection dashboard that also gathers data 952 00:34:54.250 --> 00:34:56.889 - from all these different sources that we are 953 00:34:56.889 --> 00:34:57.389 - computing 954 00:34:58.170 --> 00:34:59.069 - and that, 955 00:34:59.904 --> 00:35:02.005 - as you as you said, draw some conclusion 956 00:35:02.065 --> 00:35:03.684 - about is there a misconfiguration, 957 00:35:04.144 --> 00:35:06.944 - is the is there everything going well, which 958 00:35:06.944 --> 00:35:08.964 - is never the case in BGP, obviously. 959 00:35:10.944 --> 00:35:13.860 - But, yeah, That's, that's the main rule that 960 00:35:13.860 --> 00:35:14.880 - we're gonna do. Well in BGP. 961 00:35:15.579 --> 00:35:17.599 - Sorry? Things never go well in BGP. 962 00:35:17.980 --> 00:35:18.480 - Never. 963 00:35:19.500 --> 00:35:22.059 - Well, it's it's all fine until humans start 964 00:35:22.059 --> 00:35:23.420 - touching it, and then it becomes a huge 965 00:35:23.420 --> 00:35:23.835 - mess. 966 00:35:24.234 --> 00:35:24.734 - Exactly. 967 00:35:25.114 --> 00:35:25.614 - Yeah. 968 00:35:26.315 --> 00:35:28.175 - It works Wow. Most of the time. 969 00:35:28.635 --> 00:35:29.135 - Yeah. 970 00:35:31.434 --> 00:35:33.514 - So how would you compare and contrast this 971 00:35:33.514 --> 00:35:34.655 - to a tool like PeeringDB? 972 00:35:35.275 --> 00:35:36.635 - I mean, I know what my answer would 973 00:35:36.635 --> 00:35:38.014 - be, but I'm just curious 974 00:35:38.599 --> 00:35:41.079 - how how is that place. We're actually using 975 00:35:41.079 --> 00:35:44.039 - PillingDB also to get some, some data, like, 976 00:35:44.039 --> 00:35:46.460 - who is peeling with who. I think PillingDB, 977 00:35:46.679 --> 00:35:47.980 - it's more like a database, 978 00:35:49.159 --> 00:35:49.659 - which, 979 00:35:51.125 --> 00:35:53.364 - store, like, who is connected to who and 980 00:35:53.364 --> 00:35:55.224 - some metadata for each network, 981 00:35:55.525 --> 00:35:58.085 - but there is no actual BGP data that 982 00:35:58.085 --> 00:36:01.525 - is stored by, by by PillingDB. Right? So 983 00:36:01.525 --> 00:36:03.525 - we are really so the key difference is 984 00:36:03.525 --> 00:36:05.764 - that we are really getting raw BGP data. 985 00:36:05.764 --> 00:36:07.019 - So we have BGP sessions. 986 00:36:07.579 --> 00:36:09.980 - Getting data. We are analyzing this data, peeling 987 00:36:09.980 --> 00:36:10.480 - DB. 988 00:36:10.780 --> 00:36:11.659 - It's more like, 989 00:36:12.219 --> 00:36:15.019 - metadata, let's say, for each network. So it's 990 00:36:15.019 --> 00:36:17.179 - self reported is what it really is. Right? 991 00:36:17.179 --> 00:36:19.179 - It's also self reported. Yeah. Yeah. It's self 992 00:36:19.179 --> 00:36:21.019 - reported. It's it's what I think about my 993 00:36:21.019 --> 00:36:22.994 - network or what I know about my network. 994 00:36:23.315 --> 00:36:25.235 - Whereas what you're doing is you're actually pulling 995 00:36:25.235 --> 00:36:27.795 - the BGP tables and saying Mhmm. This is 996 00:36:27.795 --> 00:36:30.055 - what we see, and I can compare that 997 00:36:30.355 --> 00:36:31.095 - to what 998 00:36:31.715 --> 00:36:33.974 - pairing DB might happen to say. 999 00:36:34.994 --> 00:36:36.215 - Yeah. So that's interesting. 1000 00:36:37.635 --> 00:36:38.535 - So I 1001 00:36:38.969 --> 00:36:40.989 - I shouldn't think this was there anything else? 1002 00:36:41.369 --> 00:36:43.369 - I can't think of anything that I was 1003 00:36:43.369 --> 00:36:45.210 - gonna ask other than, like, okay. What's the 1004 00:36:45.210 --> 00:36:47.710 - future here? You're spending money 1005 00:36:49.289 --> 00:36:50.829 - someone's spending money 1006 00:36:51.724 --> 00:36:52.945 - storing this data. 1007 00:36:54.364 --> 00:36:55.824 - How is it being supported? 1008 00:36:56.364 --> 00:36:58.465 - What does the future look like? 1009 00:36:59.085 --> 00:37:01.664 - Is this like something people can go sponsor, 1010 00:37:01.724 --> 00:37:03.265 - companies can go sponsor? 1011 00:37:03.900 --> 00:37:06.059 - Is there a in US terms, a five 1012 00:37:06.059 --> 00:37:08.460 - zero one three c? Like, is there a 1013 00:37:08.460 --> 00:37:08.960 - nonprofit? 1014 00:37:10.219 --> 00:37:12.139 - Is there an open source project that I 1015 00:37:12.139 --> 00:37:14.940 - mean, Linux Foundation? Like, what what does this 1016 00:37:14.940 --> 00:37:15.679 - look like? 1017 00:37:16.514 --> 00:37:18.674 - So for now, it's a it's a research 1018 00:37:18.674 --> 00:37:19.174 - project, 1019 00:37:20.674 --> 00:37:22.275 - that we are doing at the University of 1020 00:37:22.275 --> 00:37:24.914 - Strasbourg. So it's paid by by, by research 1021 00:37:24.914 --> 00:37:27.234 - funding and In Germany. That that's okay. In 1022 00:37:27.234 --> 00:37:28.375 - Germany, in fact. 1023 00:37:29.155 --> 00:37:30.054 - Alright. Right. 1024 00:37:31.700 --> 00:37:33.240 - And we have a couple of sponsors. 1025 00:37:33.780 --> 00:37:35.860 - So we have the APNIC Foundation and we 1026 00:37:35.860 --> 00:37:37.240 - have OVH OVH, 1027 00:37:37.619 --> 00:37:39.300 - which is a cloud provider in France, and 1028 00:37:39.300 --> 00:37:41.539 - they are, giving us some servers on which 1029 00:37:41.539 --> 00:37:42.760 - we can run our infrastructure. 1030 00:37:43.780 --> 00:37:45.539 - Hopefully, we also have have, 1031 00:37:45.940 --> 00:37:47.525 - have worked quite hard 1032 00:37:47.984 --> 00:37:50.704 - to minimize the cost of this infrastructure, so 1033 00:37:50.704 --> 00:37:51.844 - we don't need, like, 1034 00:37:52.304 --> 00:37:55.684 - plenty of servers. Basically, with less than 1,000 1035 00:37:56.065 --> 00:37:58.224 - US dollars per month, we can we can 1036 00:37:58.224 --> 00:38:00.005 - run this stuff for now at least. 1037 00:38:00.739 --> 00:38:02.440 - So it's not so expensive, 1038 00:38:03.139 --> 00:38:04.359 - if you think about that. 1039 00:38:07.219 --> 00:38:09.539 - And, in terms of the future, well, we 1040 00:38:09.539 --> 00:38:11.159 - we hope that we can continue, 1041 00:38:12.260 --> 00:38:15.139 - operating the platform that way. But eventually, in 1042 00:38:15.139 --> 00:38:15.494 - fact, 1043 00:38:16.055 --> 00:38:18.235 - if we don't have sponsors anymore 1044 00:38:18.695 --> 00:38:20.474 - or research funding anymore, 1045 00:38:21.575 --> 00:38:23.815 - we will have either to become sort of 1046 00:38:23.815 --> 00:38:25.035 - a nonprofit organization 1047 00:38:25.974 --> 00:38:27.675 - or, or, like, a commercial, 1048 00:38:28.215 --> 00:38:28.715 - organization. 1049 00:38:29.175 --> 00:38:30.775 - We will have to to pick one of 1050 00:38:30.775 --> 00:38:31.355 - the two. 1051 00:38:32.829 --> 00:38:33.230 - And, 1052 00:38:33.710 --> 00:38:36.609 - but whatever happens whatever happens, what we really 1053 00:38:36.989 --> 00:38:39.469 - want to keep is the fact that the 1054 00:38:39.469 --> 00:38:42.609 - data we collect will remain publicly available. 1055 00:38:43.869 --> 00:38:45.389 - So that we are not going tomorrow to 1056 00:38:45.389 --> 00:38:46.844 - say, oh, we have to become commercial 1057 00:38:48.445 --> 00:38:49.565 - the data will become confidential. No. That's not 1058 00:38:49.565 --> 00:38:51.324 - what what's going to happen. So all these 1059 00:38:51.405 --> 00:38:53.244 - all the dashboards that we have, all the 1060 00:38:53.244 --> 00:38:55.585 - data we have, all of this will remain 1061 00:38:55.965 --> 00:38:58.065 - publicly available like it is now, 1062 00:38:58.525 --> 00:39:00.525 - because we want this platform to remain useful 1063 00:39:00.525 --> 00:39:01.344 - for the community. 1064 00:39:03.000 --> 00:39:05.000 - So yeah. The and and and we will 1065 00:39:05.000 --> 00:39:07.079 - have to to think about, about the funding 1066 00:39:07.079 --> 00:39:09.719 - more depending on, again, if the sponsors, they 1067 00:39:09.719 --> 00:39:12.119 - continue funding us or or or not, these 1068 00:39:12.119 --> 00:39:13.640 - kind of things. So this is still a 1069 00:39:13.640 --> 00:39:14.940 - bit unclear for now. 1070 00:39:15.714 --> 00:39:16.855 - We'll see how it goes. 1071 00:39:17.234 --> 00:39:19.714 - And, regarding the future of the platform, so 1072 00:39:19.714 --> 00:39:21.474 - we think it can evolve, well, of course, 1073 00:39:21.474 --> 00:39:22.135 - we are 1074 00:39:23.155 --> 00:39:25.554 - continuing to work quite a lot on it. 1075 00:39:25.795 --> 00:39:26.454 - We are, 1076 00:39:27.474 --> 00:39:30.179 - making, like, some more dashboards because, again, we 1077 00:39:30.179 --> 00:39:31.559 - really want to have the complete, 1078 00:39:32.019 --> 00:39:33.800 - set of tools for an operator. 1079 00:39:35.380 --> 00:39:37.800 - We also maybe thinking about maybe 1080 00:39:38.260 --> 00:39:40.900 - putting some AI into it. Know that we 1081 00:39:40.900 --> 00:39:42.280 - have all this data pipeline. 1082 00:39:42.914 --> 00:39:44.515 - Maybe there is something to do there as 1083 00:39:44.515 --> 00:39:46.914 - well. It's even some some research that we 1084 00:39:46.914 --> 00:39:48.835 - are doing here. We still like to do 1085 00:39:48.835 --> 00:39:49.335 - research, 1086 00:39:50.355 --> 00:39:51.574 - so we we are still, 1087 00:39:52.114 --> 00:39:53.735 - working on that aspect as well. 1088 00:39:54.914 --> 00:39:56.519 - And then we love BMP as well, so 1089 00:39:56.519 --> 00:39:57.719 - there is still a lot of work to 1090 00:39:57.719 --> 00:39:59.559 - do with BMP. It's it's a rather new 1091 00:39:59.559 --> 00:40:00.059 - technology. 1092 00:40:00.599 --> 00:40:01.500 - They are still, 1093 00:40:02.519 --> 00:40:03.900 - often, like, some bugs 1094 00:40:04.280 --> 00:40:05.420 - in the implementation. 1095 00:40:05.800 --> 00:40:08.840 - Yeah. So we are even yesterday, we we 1096 00:40:08.840 --> 00:40:11.074 - wrote on the burn mailing list the we 1097 00:40:11.074 --> 00:40:13.394 - reported a possible bug, so we are still, 1098 00:40:13.394 --> 00:40:15.174 - like, working a lot on that as well. 1099 00:40:15.554 --> 00:40:17.234 - We really think BNP is the is the 1100 00:40:17.234 --> 00:40:19.414 - way to go when it comes to, BGP 1101 00:40:19.474 --> 00:40:19.974 - monitoring. 1102 00:40:20.355 --> 00:40:21.875 - So we are spending a lot of time 1103 00:40:21.875 --> 00:40:22.534 - on that. 1104 00:40:24.849 --> 00:40:25.989 - Yeah. That's it. 1105 00:40:26.610 --> 00:40:27.670 - Okay. Interesting. 1106 00:40:30.210 --> 00:40:32.449 - Yeah. Cool. So I don't really I mean, 1107 00:40:32.449 --> 00:40:33.890 - I have some thoughts on by the way, 1108 00:40:33.890 --> 00:40:36.065 - you might wanna get we should might wanna 1109 00:40:36.065 --> 00:40:38.644 - start an email thread to think about, like, 1110 00:40:38.944 --> 00:40:41.025 - other places you could get funding or places 1111 00:40:41.025 --> 00:40:42.784 - where you could look for funding for this 1112 00:40:42.784 --> 00:40:44.005 - this kind of a project. 1113 00:40:44.625 --> 00:40:46.545 - But that's kind of not something we need 1114 00:40:46.545 --> 00:40:47.364 - to talk about 1115 00:40:47.905 --> 00:40:49.605 - in the podcast itself. 1116 00:40:50.180 --> 00:40:52.739 - So I don't have any more real questions. 1117 00:40:52.739 --> 00:40:54.280 - Do you have anything, Tom? 1118 00:40:55.140 --> 00:40:56.739 - No. No. This has been great, and I 1119 00:40:56.739 --> 00:40:59.539 - especially appreciate that there are three Thomas' on 1120 00:40:59.539 --> 00:41:01.059 - the call. I don't think that's ever happened. 1121 00:41:01.059 --> 00:41:02.579 - Yeah. Sorry. I guess, you know, I just 1122 00:41:02.579 --> 00:41:04.340 - said Tom, and I didn't even think about 1123 00:41:04.340 --> 00:41:04.840 - it. 1124 00:41:05.140 --> 00:41:05.640 - Yeah. 1125 00:41:06.505 --> 00:41:09.385 - The EIGRP team at Cisco used to be 1126 00:41:09.385 --> 00:41:10.284 - Tuan Tuan, 1127 00:41:11.784 --> 00:41:14.025 - Don, and Don. There are two there are 1128 00:41:14.025 --> 00:41:16.425 - two people named Tuan and two people named 1129 00:41:16.425 --> 00:41:17.304 - Don. It's 1130 00:41:18.585 --> 00:41:19.565 - it was like 1131 00:41:20.105 --> 00:41:22.789 - we always see that. It's not very original. 1132 00:41:24.450 --> 00:41:26.369 - Two dimes and two twans. And the two 1133 00:41:26.369 --> 00:41:27.809 - twans, one was a male and one was 1134 00:41:27.809 --> 00:41:29.410 - a female. So that made it even more 1135 00:41:29.410 --> 00:41:30.869 - fun. Nice. Oh, yeah. 1136 00:41:31.170 --> 00:41:33.910 - That was that was that was very entertaining. 1137 00:41:34.530 --> 00:41:35.030 - So 1138 00:41:35.410 --> 00:41:35.730 - anyway 1139 00:41:36.534 --> 00:41:37.355 - alright. Cool. 1140 00:41:39.255 --> 00:41:41.414 - So t h or t a, whichever one 1141 00:41:41.414 --> 00:41:43.655 - wants to answer, is there I assume people 1142 00:41:43.655 --> 00:41:46.295 - just go to bgproutes.io. 1143 00:41:46.295 --> 00:41:46.795 - Right? 1144 00:41:47.255 --> 00:41:49.974 - Yeah. And is there there's research papers there 1145 00:41:49.974 --> 00:41:52.359 - and other stuff you've been working on, All 1146 00:41:52.359 --> 00:41:54.199 - the stuff all the papers you've been writing 1147 00:41:54.199 --> 00:41:55.179 - are posted there 1148 00:41:55.559 --> 00:41:57.719 - in some way. Yes. They are. And we 1149 00:41:57.719 --> 00:42:00.280 - also have, like, some blog posts where we 1150 00:42:00.280 --> 00:42:02.940 - talk about, for instance, BMP or ASPA. 1151 00:42:03.639 --> 00:42:05.800 - So if you're interested in learning more about 1152 00:42:05.800 --> 00:42:06.940 - BMP, ASPA, 1153 00:42:07.744 --> 00:42:10.385 - of of course, we will add more posts, 1154 00:42:10.625 --> 00:42:13.105 - to this blog post. You can also go 1155 00:42:13.105 --> 00:42:14.144 - and read them. They are, 1156 00:42:15.025 --> 00:42:15.525 - Thomas, 1157 00:42:16.224 --> 00:42:18.864 - t h wrote them, so they are really, 1158 00:42:18.864 --> 00:42:19.605 - really interesting. 1159 00:42:22.119 --> 00:42:24.539 - Yes. We also regularly go to conferences. 1160 00:42:25.159 --> 00:42:27.000 - So I will go, for instance, to write 1161 00:42:27.000 --> 00:42:28.300 - c meeting in Belgrade, 1162 00:42:28.839 --> 00:42:31.339 - this month and also to Swedenborg in Switzerland, 1163 00:42:31.960 --> 00:42:33.739 - and we will also go to other conferences. 1164 00:42:33.880 --> 00:42:34.619 - So probably, 1165 00:42:34.934 --> 00:42:36.934 - if you're listening, maybe you will meet us 1166 00:42:36.934 --> 00:42:38.775 - at one of these conferences, and we're always 1167 00:42:38.775 --> 00:42:39.675 - happy to talk. 1168 00:42:40.135 --> 00:42:40.635 - And, 1169 00:42:41.094 --> 00:42:41.755 - of course, 1170 00:42:42.454 --> 00:42:45.094 - I everyone to connect to us or also 1171 00:42:45.094 --> 00:42:47.255 - to Right? Christian Rogers. Right? It's not just 1172 00:42:47.255 --> 00:42:48.074 - about us. 1173 00:42:51.069 --> 00:42:53.309 - Yeah. Okay. I was I was actually about 1174 00:42:53.309 --> 00:42:54.530 - to ask, so TA, 1175 00:42:55.150 --> 00:42:55.650 - LinkedIn 1176 00:42:55.949 --> 00:42:57.549 - x, like, where can people get in touch 1177 00:42:57.549 --> 00:42:59.809 - with you, follow you if they want to? 1178 00:43:00.589 --> 00:43:03.309 - So BGP runs dot I o is, present 1179 00:43:03.309 --> 00:43:06.055 - on LinkedIn. So you can follow the different 1180 00:43:06.055 --> 00:43:07.195 - news where we post 1181 00:43:07.494 --> 00:43:08.635 - our different presentations, 1182 00:43:10.055 --> 00:43:11.755 - the the blog posts. 1183 00:43:12.135 --> 00:43:14.795 - We we published several stuff, new 1184 00:43:15.255 --> 00:43:16.394 - updates on dashboards, 1185 00:43:16.839 --> 00:43:17.659 - new dashboards. 1186 00:43:18.359 --> 00:43:20.059 - And we also have, like, a Discord, 1187 00:43:20.440 --> 00:43:22.839 - where if you want to talk more directly 1188 00:43:22.839 --> 00:43:25.019 - with us, it's the best way to 1189 00:43:25.719 --> 00:43:28.219 - to do because we are pretty pretty responsive. 1190 00:43:28.855 --> 00:43:31.015 - And I understand. I I'm guessing you also 1191 00:43:31.015 --> 00:43:33.494 - have a LinkedIn account or other social media 1192 00:43:33.494 --> 00:43:35.735 - where your any other stuff you're working on, 1193 00:43:35.735 --> 00:43:37.275 - then BGP routes IO 1194 00:43:37.735 --> 00:43:40.375 - is out there as well. Right? For people 1195 00:43:40.375 --> 00:43:41.849 - to get in touch if they want to. 1196 00:43:41.930 --> 00:43:44.090 - Okay. Same thing for you. Same thing for 1197 00:43:44.090 --> 00:43:45.150 - UTH. Any 1198 00:43:45.849 --> 00:43:48.650 - do you have a blog beyond the bgproutes.io, 1199 00:43:48.650 --> 00:43:50.730 - or is it just that and then social 1200 00:43:50.730 --> 00:43:51.230 - media? 1201 00:43:52.010 --> 00:43:53.369 - You know, we are just too, so we 1202 00:43:53.369 --> 00:43:55.065 - we cannot spend too much time doing some, 1203 00:43:55.545 --> 00:43:57.885 - communication stuff. So, basically, we have a LinkedIn. 1204 00:43:58.664 --> 00:44:00.344 - You can contact us on LinkedIn. We have 1205 00:44:00.344 --> 00:44:02.105 - a Discord channel. If you want, you can 1206 00:44:02.105 --> 00:44:04.264 - also ping us there. We are and just 1207 00:44:04.264 --> 00:44:05.704 - sending us an email as well. Like, we 1208 00:44:05.704 --> 00:44:07.569 - have many operators. They want to connect with 1209 00:44:07.569 --> 00:44:09.409 - BMP. There are some bugs. They just send 1210 00:44:09.409 --> 00:44:11.250 - us an email, and we fix this together. 1211 00:44:11.250 --> 00:44:12.849 - We even sometimes set up a call with 1212 00:44:12.849 --> 00:44:14.549 - them. We had a call like that yesterday 1213 00:44:14.690 --> 00:44:17.650 - with an operator, and we could actually find 1214 00:44:17.650 --> 00:44:20.530 - the bug on birds together with this operator. 1215 00:44:20.530 --> 00:44:21.655 - So we are really, 1216 00:44:22.195 --> 00:44:23.414 - okay to just, like, 1217 00:44:23.795 --> 00:44:26.114 - have exchange a few emails, set up a 1218 00:44:26.114 --> 00:44:28.034 - quick call, and and and and make it 1219 00:44:28.034 --> 00:44:30.775 - work. So Okay. Yeah. That's it's fine. And 1220 00:44:32.034 --> 00:44:33.094 - the third time. 1221 00:44:36.339 --> 00:44:38.179 - Go ahead and say your word because it's, 1222 00:44:38.179 --> 00:44:38.579 - you know 1223 00:44:39.940 --> 00:44:40.440 - LinkedIn. 1224 00:44:41.139 --> 00:44:43.400 - LinkedIn. That's all he ever says is LinkedIn. 1225 00:44:44.019 --> 00:44:44.519 - Alright. 1226 00:44:46.500 --> 00:44:47.960 - I'm a simple man, Russ. 1227 00:44:48.579 --> 00:44:49.079 - Okay. 1228 00:44:49.565 --> 00:44:50.925 - Alright. I'm Russ White. You can find me 1229 00:44:50.925 --> 00:44:51.965 - here on the head. You can find me 1230 00:44:51.965 --> 00:44:53.644 - at rule eleven dot tech. You can find 1231 00:44:53.644 --> 00:44:55.724 - me, I don't know, here and there on 1232 00:44:55.724 --> 00:44:57.345 - different conferences and whatever, 1233 00:44:57.724 --> 00:44:59.184 - wherever I end up going. 1234 00:44:59.724 --> 00:45:01.724 - I also do have an x account routing 1235 00:45:01.724 --> 00:45:03.405 - geek. I don't really pay that much. I 1236 00:45:03.405 --> 00:45:05.760 - don't really log into it as much as 1237 00:45:05.760 --> 00:45:07.440 - I should or I don't know. Whatever you 1238 00:45:07.440 --> 00:45:08.340 - wanna say there. 1239 00:45:09.360 --> 00:45:10.800 - And we know that we live in an 1240 00:45:10.800 --> 00:45:12.559 - attention driven economy, so we thank you for 1241 00:45:12.559 --> 00:45:14.079 - listening all the way to the bitter end 1242 00:45:14.079 --> 00:45:15.920 - of this episode of The Hedge, and we 1243 00:45:15.920 --> 00:45:17.300 - will catch you next time. 1244 00:45:24.844 --> 00:45:25.344 - Time.