1 00:00:01,599 --> 00:00:03,600 Join us as we gather around the hedge, 2 00:00:03,600 --> 00:00:06,580 where we dig into technology, business, and culture 3 00:00:06,639 --> 00:00:08,980 with the finest minds in computer networking. 4 00:00:20,574 --> 00:00:21,474 Hello, Tom. 5 00:00:22,494 --> 00:00:23,634 Hello, Russ. 6 00:00:24,974 --> 00:00:27,375 The plant is back and looking healthy. Is 7 00:00:27,375 --> 00:00:29,489 that a new plant? No. It's back. It's 8 00:00:29,489 --> 00:00:30,929 looking healthy, and I got a lot more 9 00:00:30,929 --> 00:00:32,210 light in this room now because of the 10 00:00:32,210 --> 00:00:34,049 rearrangement. So I think the plant's happy with 11 00:00:34,049 --> 00:00:36,149 me. You sent it to the plant hospital 12 00:00:36,530 --> 00:00:36,770 and 13 00:00:37,649 --> 00:00:40,070 Yeah. It convalesced there and then came back. 14 00:00:42,765 --> 00:00:45,564 And today, we are joined by with Jeff 15 00:00:45,564 --> 00:00:46,784 or by Jeff Houston. 16 00:00:47,244 --> 00:00:50,125 So, Jeff, are you in Australia with your 17 00:00:50,125 --> 00:00:51,984 bears and and and 18 00:00:53,405 --> 00:00:53,905 and 19 00:00:54,604 --> 00:00:54,684 and 20 00:00:55,929 --> 00:00:58,250 Various bits of native wildlife. Yes. I'm in 21 00:00:58,250 --> 00:01:00,329 Australia at the moment, at the height of 22 00:01:00,329 --> 00:01:00,829 summer. 23 00:01:01,929 --> 00:01:04,009 So, yes, it's it's it's a it's a 24 00:01:04,009 --> 00:01:06,269 wonderful time at that point because, you know, 25 00:01:06,569 --> 00:01:08,189 winter's winter. But, 26 00:01:11,094 --> 00:01:12,155 what can I say? 27 00:01:12,614 --> 00:01:15,034 Yes. Yes. Yes. We all know about winter. 28 00:01:15,094 --> 00:01:16,635 Well, we all know about Australia. 29 00:01:17,015 --> 00:01:18,534 We've all heard it's a great place to 30 00:01:18,534 --> 00:01:20,635 visit as long as you avoid the wildlife. 31 00:01:24,939 --> 00:01:27,420 Everyone's got a week. Just remember that. Everyone's 32 00:01:27,420 --> 00:01:28,239 got a week. 33 00:01:29,979 --> 00:01:31,599 Even the even the crocs. 34 00:01:31,979 --> 00:01:33,759 I have seen the videos 35 00:01:34,299 --> 00:01:35,280 of the kangaroos 36 00:01:35,819 --> 00:01:37,359 beating up on people. Okay? 37 00:01:40,864 --> 00:01:42,864 Yeah. Well, they're not cute and cuddly. Just 38 00:01:42,864 --> 00:01:43,685 remember that. 39 00:01:46,625 --> 00:01:48,545 So today, we're talking about v four address 40 00:01:48,545 --> 00:01:50,704 of the BGP table. Right? The BGP table. 41 00:01:50,704 --> 00:01:52,564 And what's going on with the BGP table? 42 00:01:52,864 --> 00:01:55,530 Alright. So talk to me. Has the BGP 43 00:01:55,530 --> 00:01:57,549 table grown? Is it smaller? 44 00:01:58,329 --> 00:02:00,270 Has it, like, exponential? 45 00:02:00,969 --> 00:02:01,630 Is it 46 00:02:02,090 --> 00:02:02,829 more flappy? 47 00:02:03,210 --> 00:02:03,710 Is 48 00:02:04,490 --> 00:02:05,390 it more flappy? 49 00:02:06,555 --> 00:02:08,814 The thing to remember about the BGP table 50 00:02:09,275 --> 00:02:11,055 is that no one's in charge. 51 00:02:11,914 --> 00:02:13,134 Absolutely no one. 52 00:02:13,435 --> 00:02:15,754 And when you wanna sort of join the 53 00:02:15,754 --> 00:02:18,394 Internet or add more things into the Internet 54 00:02:18,394 --> 00:02:19,294 in terms of reachability, 55 00:02:19,995 --> 00:02:20,655 no permission. 56 00:02:21,000 --> 00:02:22,379 Just do it. 57 00:02:23,319 --> 00:02:25,500 Right? So you've got this system 58 00:02:25,879 --> 00:02:27,979 where everyone has to carry 59 00:02:28,280 --> 00:02:31,500 BGP tables. Everyone has to carry that information, 60 00:02:32,120 --> 00:02:33,659 but there's no gatekeeper 61 00:02:34,120 --> 00:02:35,180 on getting there. 62 00:02:35,705 --> 00:02:36,685 It's a completely 63 00:02:36,985 --> 00:02:37,485 unconstrained 64 00:02:37,864 --> 00:02:38,364 system. 65 00:02:39,384 --> 00:02:39,884 So 66 00:02:40,185 --> 00:02:42,044 when you buy your next router 67 00:02:42,424 --> 00:02:44,104 and you go, well, it's gotta last for 68 00:02:44,104 --> 00:02:45,544 at least five years because it was a 69 00:02:45,544 --> 00:02:47,784 big expensive beastie, and I paid some vendor 70 00:02:47,784 --> 00:02:49,864 heaps of money, blah blah blah, then then 71 00:02:49,864 --> 00:02:51,719 the issue is, how big? 72 00:02:53,060 --> 00:02:53,879 Do I need 73 00:02:54,900 --> 00:02:57,540 a million entries in the routing table? Do 74 00:02:57,540 --> 00:03:00,099 I need 2,000,000? Do I need 10,000,000? Now 75 00:03:00,099 --> 00:03:02,840 the thing to remember about this is that 76 00:03:03,504 --> 00:03:04,965 I suspect that 77 00:03:05,425 --> 00:03:08,305 the router line card memory is the most 78 00:03:08,305 --> 00:03:10,324 expensive memory you could buy. 79 00:03:11,185 --> 00:03:13,365 Because as we keep on upping our transmission 80 00:03:13,504 --> 00:03:15,425 speeds, and, you know, I've just heard someone 81 00:03:15,425 --> 00:03:16,884 has done terabit fibre, 82 00:03:17,344 --> 00:03:20,200 then that's an awful lot of packets per 83 00:03:20,200 --> 00:03:20,700 second. 84 00:03:21,159 --> 00:03:23,480 Which means an awful lot of silicon working 85 00:03:23,480 --> 00:03:24,540 very, very hard 86 00:03:24,840 --> 00:03:27,639 to look up the destination address of these 87 00:03:27,639 --> 00:03:28,139 packets 88 00:03:28,599 --> 00:03:29,819 for every packet. 89 00:03:31,514 --> 00:03:33,674 Now, if you've got a million entries in 90 00:03:33,674 --> 00:03:34,814 your routing table, 91 00:03:35,354 --> 00:03:37,674 no matter how you pack that up, that's 92 00:03:37,674 --> 00:03:39,294 going to be 20 lookups. 93 00:03:40,074 --> 00:03:41,834 So in any kind of binary chop in 94 00:03:41,834 --> 00:03:44,155 the most efficient mechanism, it might well be 95 00:03:44,155 --> 00:03:44,655 more. 96 00:03:45,450 --> 00:03:47,770 That's a lot of work when you're trying 97 00:03:47,770 --> 00:03:50,090 to shovel through a few hundred million packets 98 00:03:50,090 --> 00:03:50,830 per second. 99 00:03:51,450 --> 00:03:51,950 So 100 00:03:52,330 --> 00:03:53,790 the size of the table 101 00:03:54,490 --> 00:03:56,810 is actually crucially important. And it's not just 102 00:03:56,810 --> 00:03:58,590 the size of the table today. 103 00:03:59,210 --> 00:04:00,670 It's the size of the table 104 00:04:01,194 --> 00:04:03,034 near the end of the lifetime of the 105 00:04:03,034 --> 00:04:04,415 router you've just bought. 106 00:04:04,794 --> 00:04:06,395 Because if you buy a router to cope 107 00:04:06,395 --> 00:04:07,135 with today, 108 00:04:08,155 --> 00:04:10,635 if the table's growing, you've got a problem 109 00:04:10,635 --> 00:04:11,135 tomorrow. 110 00:04:12,474 --> 00:04:14,715 And so looking at the size of the 111 00:04:14,715 --> 00:04:17,330 BGP routing table and its dynamic behaviour 112 00:04:17,949 --> 00:04:19,970 is actually critically important for 113 00:04:20,270 --> 00:04:22,449 literally everyone who routes the internet. 114 00:04:23,230 --> 00:04:25,410 Okay? So not a small question. 115 00:04:26,029 --> 00:04:28,430 Very big question if your job is pushing 116 00:04:28,430 --> 00:04:28,930 packets. 117 00:04:29,545 --> 00:04:31,384 And by the way, line card memory is 118 00:04:31,384 --> 00:04:32,045 not just 119 00:04:32,985 --> 00:04:35,145 in terms of entries. It's also length of 120 00:04:35,145 --> 00:04:37,324 entry and a lot of forwarding tables 121 00:04:38,264 --> 00:04:40,365 and a lot of forwarding engines. Because 122 00:04:41,384 --> 00:04:43,625 longer entries take up more space, blah, blah, 123 00:04:43,625 --> 00:04:44,125 blah, 124 00:04:44,459 --> 00:04:46,399 types of things that you can run into. 125 00:04:47,019 --> 00:04:49,740 And so, yeah. It's The disaster that is 126 00:04:49,740 --> 00:04:51,660 v six. Yes. That's what I was gonna 127 00:04:51,660 --> 00:04:53,899 say. Yes. V six is where this becomes 128 00:04:53,899 --> 00:04:56,720 an issue. And so therefore, yeah, you you 129 00:04:57,660 --> 00:05:00,525 it's not just it's also rate of change 130 00:05:00,525 --> 00:05:02,205 and all those other things like that as 131 00:05:02,205 --> 00:05:02,705 well. 132 00:05:04,205 --> 00:05:05,884 Yeah. We we can get on to that. 133 00:05:05,884 --> 00:05:06,384 So 134 00:05:07,004 --> 00:05:09,645 let's just talk quickly about v four. And 135 00:05:09,645 --> 00:05:12,285 the issue about v four is it was 136 00:05:12,285 --> 00:05:12,785 tracking 137 00:05:13,379 --> 00:05:17,060 an exponential trend accrue through much of the 138 00:05:17,060 --> 00:05:18,600 naughties of the February. 139 00:05:18,900 --> 00:05:20,980 It was the expansion of the internet into 140 00:05:20,980 --> 00:05:23,460 true mass market retail. And, you know, the 141 00:05:23,460 --> 00:05:25,220 line was indeed up and to the right 142 00:05:25,220 --> 00:05:26,745 and growing pretty quickly. 143 00:05:27,544 --> 00:05:29,785 But at the same time, the pool of 144 00:05:29,785 --> 00:05:30,285 addresses 145 00:05:30,985 --> 00:05:33,084 that we were using to feed that growth 146 00:05:33,464 --> 00:05:35,305 was reaching its end. This is the whole 147 00:05:35,305 --> 00:05:37,464 reason about v six. We predicted it back 148 00:05:37,464 --> 00:05:39,790 in 1990 that we'd run out and, you 149 00:05:39,790 --> 00:05:41,470 know, true to form, a bit later than 150 00:05:41,470 --> 00:05:42,689 we had thought, but nevertheless, 151 00:05:43,069 --> 00:05:44,370 we ran out. 152 00:05:45,149 --> 00:05:47,330 Now, if this was the telephone network 153 00:05:47,870 --> 00:05:50,270 of days gone by, when you ran out 154 00:05:50,270 --> 00:05:52,290 of telephone numbers, you ran out of telephones 155 00:05:52,350 --> 00:05:53,009 to connect. 156 00:05:53,814 --> 00:05:57,415 Simple as that. But in the internet, you've 157 00:05:57,415 --> 00:05:59,894 kind of got these computers all over the 158 00:05:59,894 --> 00:06:00,394 place 159 00:06:00,694 --> 00:06:02,875 who are able to do clever things. 160 00:06:03,654 --> 00:06:04,555 Share addresses. 161 00:06:05,415 --> 00:06:07,189 Fake the fact that you haven't got enough 162 00:06:07,189 --> 00:06:09,770 by sharing them around, sharing them over time, 163 00:06:09,910 --> 00:06:12,150 and even invading some of the, what we 164 00:06:12,150 --> 00:06:14,389 call, the port number space in TCP and 165 00:06:14,389 --> 00:06:14,889 UDP, 166 00:06:15,750 --> 00:06:17,129 and sort of squeezing 167 00:06:17,590 --> 00:06:20,389 10 or even 20 computers onto one IP 168 00:06:20,389 --> 00:06:20,889 address 169 00:06:21,675 --> 00:06:23,834 by by sharing out some pools of port 170 00:06:23,834 --> 00:06:25,214 numbers to each of them. 171 00:06:26,074 --> 00:06:27,995 And so even though we were running out 172 00:06:27,995 --> 00:06:31,194 in the in February and 02/2010 173 00:06:31,194 --> 00:06:31,855 and onward, 174 00:06:32,714 --> 00:06:34,975 the growth of the Internet didn't stop. 175 00:06:35,439 --> 00:06:36,259 Just didn't. 176 00:06:36,720 --> 00:06:39,439 We're running on empty, but no one seemed 177 00:06:39,439 --> 00:06:40,579 to know or care. 178 00:06:41,360 --> 00:06:44,579 And this bizarre situation carried on for, 179 00:06:44,879 --> 00:06:47,459 jeez, right up until around 2021. 180 00:06:48,095 --> 00:06:48,995 Just growth. 181 00:06:49,535 --> 00:06:52,014 At the time when the central pools ran 182 00:06:52,014 --> 00:06:53,555 out, we had about, oh, 183 00:06:54,735 --> 00:06:55,694 100,000 184 00:06:55,694 --> 00:06:57,154 entries in the routing table. 185 00:06:58,414 --> 00:07:00,240 By the time things are getting a bit 186 00:07:00,399 --> 00:07:03,759 tight in 2021, we're having 950,000 187 00:07:03,759 --> 00:07:05,300 entries in the routing table. 188 00:07:06,000 --> 00:07:06,500 Wow. 189 00:07:07,039 --> 00:07:08,899 What a collective act of juggling. 190 00:07:09,919 --> 00:07:11,759 And it was only really in the last 191 00:07:11,759 --> 00:07:12,579 three years 192 00:07:13,279 --> 00:07:13,779 that 193 00:07:14,514 --> 00:07:16,854 the pace of growth has slackened off. 194 00:07:17,314 --> 00:07:19,574 And instead of rising by, you know, 195 00:07:19,875 --> 00:07:22,274 60 to a hundred thousand entries per year, 196 00:07:22,274 --> 00:07:24,274 which is where it was heading, the growth 197 00:07:24,274 --> 00:07:25,894 rate's now a lot, lot slower. 198 00:07:26,939 --> 00:07:27,439 It's 199 00:07:27,980 --> 00:07:29,180 around 40,000 200 00:07:29,180 --> 00:07:30,400 entries per year. 201 00:07:30,779 --> 00:07:33,100 It's still rising. Absolutely. And we're now at 202 00:07:33,100 --> 00:07:35,279 about a million entries in the routing table. 203 00:07:35,740 --> 00:07:38,080 But that frenetic pace of growth 204 00:07:38,540 --> 00:07:39,920 is slackening off. 205 00:07:40,354 --> 00:07:42,834 We're doing around about a hundred new entries 206 00:07:42,834 --> 00:07:45,254 a day in the v four routing table. 207 00:07:48,435 --> 00:07:50,294 So is that a problem? 208 00:07:52,449 --> 00:07:53,189 Good question. 209 00:07:53,810 --> 00:07:54,949 Yeah. Could be. 210 00:07:55,330 --> 00:07:56,310 Could be. Right? 211 00:07:56,930 --> 00:07:59,730 You're juggling you're juggling around two things. And 212 00:07:59,730 --> 00:08:01,430 and one thing is Moore's law, 213 00:08:02,290 --> 00:08:05,029 where we're able, or have been so far, 214 00:08:05,324 --> 00:08:07,725 to cram twice as many gates into a 215 00:08:07,725 --> 00:08:10,604 silicon chip about every eighteen months and do 216 00:08:10,604 --> 00:08:12,305 so at pretty constant price. 217 00:08:13,564 --> 00:08:16,125 So if the cost of routing is, you 218 00:08:16,125 --> 00:08:16,625 know, 219 00:08:17,164 --> 00:08:20,704 $1 per unit of of of routing table, 220 00:08:21,220 --> 00:08:23,220 then as long as your growth is is 221 00:08:23,220 --> 00:08:25,399 less than doubling every 18, 222 00:08:26,419 --> 00:08:28,899 theoretically, there are chips around in eighteen months' 223 00:08:28,899 --> 00:08:30,819 time that have double the capacity at the 224 00:08:30,819 --> 00:08:31,560 same price. 225 00:08:32,580 --> 00:08:33,080 Theoretically. 226 00:08:33,965 --> 00:08:35,644 I was gonna say Moore's law is dead. 227 00:08:35,644 --> 00:08:37,404 Right? I mean, that's what I'm dealing with. 228 00:08:37,404 --> 00:08:39,425 Moore's law is completely dead. Is dead. 229 00:08:40,125 --> 00:08:41,585 We're we're down at 230 00:08:41,965 --> 00:08:43,425 at small numbers of nanometers, 231 00:08:44,205 --> 00:08:45,345 three, four, five. 232 00:08:45,819 --> 00:08:47,419 We don't know how to get to one. 233 00:08:47,419 --> 00:08:49,019 It's kind of the width of a silicon 234 00:08:49,019 --> 00:08:51,579 atom. This is this is physics is getting 235 00:08:51,579 --> 00:08:54,299 in the way, and and further prodigious growth 236 00:08:54,299 --> 00:08:55,600 is kind of not. 237 00:08:56,220 --> 00:08:58,754 And what it also means is you can't 238 00:08:58,754 --> 00:09:01,235 just count on the future to solve today's 239 00:09:01,235 --> 00:09:03,875 problem if you do nothing. Moore's Law won't 240 00:09:03,875 --> 00:09:04,934 come to your rescue. 241 00:09:05,715 --> 00:09:07,495 That's kind of the first problem. 242 00:09:08,195 --> 00:09:11,075 The second problem, though, is that we've yet 243 00:09:11,075 --> 00:09:12,915 to actually figure out how far we can 244 00:09:12,915 --> 00:09:13,894 push photons. 245 00:09:15,259 --> 00:09:16,000 And while 246 00:09:16,459 --> 00:09:18,399 we're getting fiber optic systems 247 00:09:19,019 --> 00:09:21,839 that still increase in channel capacity, 248 00:09:22,939 --> 00:09:25,019 we've got this underlying issue of, well, what 249 00:09:25,019 --> 00:09:26,459 do you do with the silicon at either 250 00:09:26,459 --> 00:09:27,439 end of the fiber? 251 00:09:28,705 --> 00:09:29,205 Because 252 00:09:29,745 --> 00:09:32,705 so far, we've done extraordinary things in digital 253 00:09:32,705 --> 00:09:35,445 signal processing, and it's not stopping yet. 254 00:09:36,065 --> 00:09:36,565 Yet. 255 00:09:37,264 --> 00:09:39,504 And, you know, the latest announcement, one terabit 256 00:09:39,504 --> 00:09:42,370 per channel. And that's kind of, wow. That's 257 00:09:42,370 --> 00:09:44,709 a huge amount of packets hitting the silicon. 258 00:09:46,209 --> 00:09:48,230 So it's the Now, now, is this silicon 259 00:09:48,289 --> 00:09:48,789 photonics, 260 00:09:49,409 --> 00:09:51,269 or is this still electrical interface? 261 00:09:52,769 --> 00:09:55,945 Digital signal processing. It's trying to extract the 262 00:09:55,945 --> 00:09:57,964 digital signal out of the photonics. 263 00:09:58,745 --> 00:10:01,704 Okay. Right. Okay. And there's a relationship between, 264 00:10:01,704 --> 00:10:03,325 if you will, how good it is, 265 00:10:04,264 --> 00:10:06,904 versus the track width on on the underlying 266 00:10:06,904 --> 00:10:09,059 chip. What you've gotta do is 267 00:10:09,539 --> 00:10:11,700 to get more capacity out of a fiber 268 00:10:11,700 --> 00:10:14,179 signal, you've actually got to put more points 269 00:10:14,179 --> 00:10:17,720 in phase amplitude coding space, in Fourier space. 270 00:10:18,419 --> 00:10:21,620 And the sensitivity of detecting that actually depends 271 00:10:21,620 --> 00:10:24,565 on how good your signal processes are, which 272 00:10:24,565 --> 00:10:26,725 depends on how many gates you can stuff 273 00:10:26,725 --> 00:10:29,524 in a square centimeter of silicon, basically. Right. 274 00:10:29,524 --> 00:10:31,865 Yeah. You know, the two are directly related. 275 00:10:32,565 --> 00:10:34,264 What about what about the growth? 276 00:10:35,284 --> 00:10:36,884 Do you have a sense of are we 277 00:10:36,884 --> 00:10:38,565 getting are we getting short are we getting 278 00:10:38,565 --> 00:10:40,620 longer prefixes announced now if we're up to 279 00:10:40,620 --> 00:10:41,980 a hundred if we're down to a hundred 280 00:10:41,980 --> 00:10:44,540 a day, or what's the what's the size 281 00:10:44,540 --> 00:10:45,920 distribution looking like? 282 00:10:47,340 --> 00:10:49,840 That's a linear increase, and and linear's 283 00:10:50,220 --> 00:10:51,279 kind of good, 284 00:10:51,980 --> 00:10:52,480 kinda. 285 00:10:54,504 --> 00:10:56,584 Just depends on how long it lasts and 286 00:10:56,584 --> 00:10:57,784 what the rate of lending is. It does 287 00:10:57,784 --> 00:10:59,704 depend on how long it lasts. But but, 288 00:10:59,704 --> 00:11:02,264 in essence, what's going on right now is 289 00:11:02,264 --> 00:11:04,845 that the average announcement size in in BGP 290 00:11:05,384 --> 00:11:08,204 is still around 3,200 291 00:11:08,345 --> 00:11:09,164 IP addresses. 292 00:11:09,799 --> 00:11:10,620 So it's 293 00:11:11,639 --> 00:11:13,019 slash 20 or so. 294 00:11:13,559 --> 00:11:16,039 We haven't we haven't got to everything's a 295 00:11:16,039 --> 00:11:17,100 slash 24. 296 00:11:17,799 --> 00:11:19,959 If we did, we'd have 4,000,000,000 entries in 297 00:11:19,959 --> 00:11:21,419 the routing table, and 298 00:11:21,875 --> 00:11:24,914 things would melt. You know, 1,000,000 is still 299 00:11:24,914 --> 00:11:25,735 pretty low 300 00:11:26,274 --> 00:11:27,575 in the scheme of things. 301 00:11:29,475 --> 00:11:31,095 The interesting news in v4 302 00:11:32,034 --> 00:11:33,954 is I think it's time to start talking 303 00:11:33,954 --> 00:11:36,375 about the saturation word, the s word. 304 00:11:37,460 --> 00:11:38,840 The pace of growth 305 00:11:39,460 --> 00:11:40,200 is actually 306 00:11:41,059 --> 00:11:42,120 slackening off, 307 00:11:42,660 --> 00:11:43,160 that 308 00:11:43,700 --> 00:11:47,320 the previous heady days of 200 new entries 309 00:11:47,379 --> 00:11:48,120 per day 310 00:11:48,500 --> 00:11:49,240 and more 311 00:11:49,779 --> 00:11:51,320 is now pretty much gone. 312 00:11:51,934 --> 00:11:53,774 And while there are certainly some days that 313 00:11:53,774 --> 00:11:55,554 are exceptions in the routing table, 314 00:11:56,095 --> 00:11:57,875 we're 100 and falling. 315 00:11:58,894 --> 00:12:01,375 A hundred and falling. And that's the number 316 00:12:01,375 --> 00:12:02,115 of ASs. 317 00:12:02,735 --> 00:12:04,034 It's the pace of growth. 318 00:12:04,815 --> 00:12:05,460 The big 319 00:12:05,779 --> 00:12:06,840 Asian powerhouses 320 00:12:07,379 --> 00:12:08,679 of v4 growth 321 00:12:08,980 --> 00:12:09,800 have stopped. 322 00:12:10,420 --> 00:12:13,540 By and large, they're substituting v six wherever 323 00:12:13,540 --> 00:12:15,960 they can because, you know, it's cheaper, 324 00:12:17,540 --> 00:12:18,980 because you don't have to work on the 325 00:12:18,980 --> 00:12:19,480 market 326 00:12:20,165 --> 00:12:21,144 to buy addresses, 327 00:12:21,605 --> 00:12:25,465 v4 addresses at still what is kicking around 328 00:12:26,004 --> 00:12:27,945 $30 to $50 an address. 329 00:12:28,644 --> 00:12:31,225 V six is nowhere near that cost. So 330 00:12:32,085 --> 00:12:33,945 in some ways, the growth factors 331 00:12:34,730 --> 00:12:36,350 in v four are coming down. 332 00:12:37,370 --> 00:12:40,269 It seems that the best of the mathematical 333 00:12:40,409 --> 00:12:42,250 projections in v four, if you try and 334 00:12:42,250 --> 00:12:44,029 curve fit the last five years, 335 00:12:44,970 --> 00:12:47,769 tend to suggest that we're close to peak 336 00:12:47,769 --> 00:12:48,509 v four. 337 00:12:50,745 --> 00:12:53,004 Around about a million entries. And 338 00:12:53,625 --> 00:12:55,325 if those mathematical projections 339 00:12:56,024 --> 00:12:58,985 have any sort of backing to it, that's 340 00:12:58,985 --> 00:13:00,825 where it's going to sit for some years 341 00:13:00,825 --> 00:13:02,764 and then possibly decline. 342 00:13:03,384 --> 00:13:04,924 But that's just maths. 343 00:13:05,919 --> 00:13:06,740 The sociological 344 00:13:07,360 --> 00:13:08,419 sort of pressure is 345 00:13:08,879 --> 00:13:11,039 once it's viable to run v six only 346 00:13:11,039 --> 00:13:11,539 networks, 347 00:13:13,279 --> 00:13:15,199 v four has a real problem. And it's 348 00:13:15,199 --> 00:13:17,220 more like a cliff than a gentle decline 349 00:13:17,600 --> 00:13:18,659 once it's viable. 350 00:13:19,495 --> 00:13:20,954 And part of the reason is 351 00:13:21,654 --> 00:13:22,475 we're now 352 00:13:23,334 --> 00:13:25,754 way inside scarcity price premiums, 353 00:13:26,855 --> 00:13:27,995 scarcity costs. 354 00:13:28,774 --> 00:13:31,274 And that's what's driving, if you will, 355 00:13:31,919 --> 00:13:32,980 this thing about, 356 00:13:33,679 --> 00:13:35,200 well, if I've got a problem and I 357 00:13:35,200 --> 00:13:37,839 need to expand a network right now here 358 00:13:37,839 --> 00:13:38,339 today, 359 00:13:39,440 --> 00:13:41,839 I can buy more v six addresses sorry, 360 00:13:41,839 --> 00:13:43,519 v four addresses, or I can get some 361 00:13:43,519 --> 00:13:44,179 v six. 362 00:13:45,985 --> 00:13:47,764 The only thing that's happening 363 00:13:48,225 --> 00:13:49,605 in the v4 space 364 00:13:50,065 --> 00:13:51,204 are actually clouds, 365 00:13:52,625 --> 00:13:54,164 the only thing that's clouds. 366 00:13:54,784 --> 00:13:55,284 And 367 00:13:55,985 --> 00:13:56,644 the last 368 00:13:57,424 --> 00:13:57,924 big, 369 00:13:58,279 --> 00:14:00,940 big increase in the amount of addresses advertised 370 00:14:01,960 --> 00:14:02,700 in the 371 00:14:03,720 --> 00:14:04,620 routing space, 372 00:14:05,639 --> 00:14:09,000 the size, if you will, expressed in number 373 00:14:09,000 --> 00:14:11,019 of entries in the routing table, 374 00:14:11,784 --> 00:14:14,044 is actually a massive increase by Amazon. 375 00:14:14,985 --> 00:14:16,504 Bill, on the, I think it was the 376 00:14:16,504 --> 00:14:17,565 December 12, 377 00:14:18,985 --> 00:14:20,445 pumped up another, you know, 378 00:14:22,904 --> 00:14:23,965 80,000,000 379 00:14:24,105 --> 00:14:24,605 addresses. 380 00:14:25,879 --> 00:14:28,139 So it's not the clients, it's the clouds. 381 00:14:28,679 --> 00:14:30,620 It's the other side. And interestingly, 382 00:14:31,000 --> 00:14:32,299 it's only Amazon. 383 00:14:32,840 --> 00:14:35,019 It's not everyone else in the cloud business. 384 00:14:35,720 --> 00:14:37,320 Lots of the others who are in the 385 00:14:37,320 --> 00:14:39,160 cloud business are doing it on v four 386 00:14:39,160 --> 00:14:39,899 with NATS. 387 00:14:40,315 --> 00:14:42,715 But for some reason, Amazon has sort of 388 00:14:42,715 --> 00:14:44,955 pulled together a large amount of addresses and 389 00:14:44,955 --> 00:14:47,455 and bumped it into the routing table, 390 00:14:47,995 --> 00:14:50,715 as v four connectivity, which is, you know, 391 00:14:50,715 --> 00:14:52,715 unusual, but they know what they're doing, I'm 392 00:14:52,715 --> 00:14:53,215 sure. 393 00:14:56,610 --> 00:14:57,990 Interesting. So so 394 00:14:58,610 --> 00:15:00,529 yeah. So that means that we're gonna be 395 00:15:00,529 --> 00:15:02,929 watching the v four space not continue to 396 00:15:02,929 --> 00:15:03,429 grow, 397 00:15:04,129 --> 00:15:05,889 but the v six space is gonna start 398 00:15:05,970 --> 00:15:08,450 v six. Yeah? Yes. Which means v six 399 00:15:08,450 --> 00:15:10,665 space is gonna take off, which is not 400 00:15:10,665 --> 00:15:13,245 necessarily a good thing because I don't know. 401 00:15:13,304 --> 00:15:15,465 You know, the average what what is the 402 00:15:15,465 --> 00:15:18,285 average advertisement length in v six right now? 403 00:15:18,424 --> 00:15:19,725 Slash 32. 404 00:15:20,585 --> 00:15:23,070 Okay. So two to the thirty second is? 405 00:15:23,789 --> 00:15:24,529 4,000,000,000. 406 00:15:25,470 --> 00:15:26,450 Yeah. So 407 00:15:27,389 --> 00:15:28,450 4,000,000,000 advertisements. 408 00:15:30,429 --> 00:15:33,710 Actually, it it's kinda weird because the v 409 00:15:33,710 --> 00:15:35,090 six phase is almost 410 00:15:36,029 --> 00:15:37,169 split in half. 411 00:15:37,634 --> 00:15:41,095 The most common prefix is a slash 48. 412 00:15:42,514 --> 00:15:44,274 And if you care to do your maths, 413 00:15:44,274 --> 00:15:46,035 two to the 48 is a very, very 414 00:15:46,035 --> 00:15:46,774 big number, 415 00:15:47,634 --> 00:15:50,215 bigger than, you know, we can cope with. 416 00:15:50,675 --> 00:15:53,740 And so if v six took off and 417 00:15:53,740 --> 00:15:56,320 everyone has their own v six address and 418 00:15:56,940 --> 00:15:59,019 everyone tries to separately announce it in the 419 00:15:59,019 --> 00:16:02,300 routing table, you know, ain't gonna work. Just 420 00:16:02,300 --> 00:16:02,800 ain't. 421 00:16:03,740 --> 00:16:06,540 The v six routing table has a similar 422 00:16:06,540 --> 00:16:07,279 sort of 423 00:16:07,705 --> 00:16:09,725 life and profile to v4, 424 00:16:10,424 --> 00:16:12,825 except it's sort of pushed ten years in 425 00:16:12,825 --> 00:16:13,485 the future. 426 00:16:14,264 --> 00:16:16,745 Where the early days of the internet and 427 00:16:16,745 --> 00:16:19,304 v4 were exponential growth through the '90s and 428 00:16:19,304 --> 00:16:20,125 early '00s, 429 00:16:20,920 --> 00:16:21,740 v six, 430 00:16:22,279 --> 00:16:25,019 that same exponential growth has occurred 431 00:16:25,800 --> 00:16:26,860 in the 2010s. 432 00:16:29,000 --> 00:16:31,980 And it sort of ramped up. But interestingly, 433 00:16:32,519 --> 00:16:34,700 the same slackening off 434 00:16:35,485 --> 00:16:38,045 that we saw in v four, we actually 435 00:16:38,045 --> 00:16:39,105 see in v six. 436 00:16:39,804 --> 00:16:41,504 And even more interestingly, 437 00:16:43,644 --> 00:16:46,365 the relative growth, v four, it's come down 438 00:16:46,365 --> 00:16:48,205 from 200 a day to a hundred a 439 00:16:48,205 --> 00:16:50,899 day, new entries per day. In v six, 440 00:16:50,899 --> 00:16:53,860 it never got to 200 a day except 441 00:16:53,860 --> 00:16:55,079 for one day 442 00:16:55,459 --> 00:16:57,159 in 02/2001. 443 00:16:58,740 --> 00:17:01,159 It it's always been, even this massive exponential 444 00:17:01,220 --> 00:17:03,595 growth of v six. It was growing from 445 00:17:03,595 --> 00:17:05,835 50 entries a day, and this, you know, 446 00:17:05,835 --> 00:17:08,394 finally got to 200. Immediately afterwards, it came 447 00:17:08,394 --> 00:17:09,134 down again. 448 00:17:09,515 --> 00:17:11,375 It's currently growing at about 449 00:17:11,674 --> 00:17:14,015 50 to 60 new entries a day. 450 00:17:14,394 --> 00:17:16,714 So the growth rate's actually slower than v 451 00:17:16,714 --> 00:17:17,214 four. 452 00:17:17,835 --> 00:17:20,069 I wonder why that is given the pressure 453 00:17:21,409 --> 00:17:23,890 on as as the pressure comes off v 454 00:17:23,890 --> 00:17:25,890 four, why? Is it just because these 455 00:17:26,529 --> 00:17:28,210 when you look at a slash 32, it 456 00:17:28,210 --> 00:17:30,130 contains so much that you don't have to 457 00:17:30,130 --> 00:17:32,914 advertise very many of them to cover lots 458 00:17:32,914 --> 00:17:35,575 of stuff? We've given every person 459 00:17:36,515 --> 00:17:39,075 a a computer device. We've given them five. 460 00:17:39,075 --> 00:17:41,414 We're even putting SIM cards in watches. 461 00:17:43,474 --> 00:17:45,394 So where's it gonna grow? Yeah. Don't don't 462 00:17:45,394 --> 00:17:46,140 David. Like 463 00:17:46,700 --> 00:17:47,920 Now, where's the growth? 464 00:17:49,019 --> 00:17:49,680 You know? 465 00:17:50,140 --> 00:17:52,140 At at some point, you've gotta look at 466 00:17:52,140 --> 00:17:55,180 the underlying market dynamics and really do ask 467 00:17:55,180 --> 00:17:55,680 yourself, 468 00:17:56,700 --> 00:17:57,839 where is the next 469 00:17:58,140 --> 00:17:59,839 double in size for the Internet? 470 00:18:00,445 --> 00:18:03,105 It isn't devices you carry because 471 00:18:03,404 --> 00:18:05,565 they're too expensive to double in size because 472 00:18:05,565 --> 00:18:08,525 everyone's got one. It's not cars. It's not 473 00:18:08,525 --> 00:18:09,025 watches. 474 00:18:09,484 --> 00:18:11,904 Certainly not, you know, stand alone computers. 475 00:18:13,660 --> 00:18:16,299 It's not even household devices because we netted 476 00:18:16,299 --> 00:18:16,799 them. 477 00:18:17,740 --> 00:18:18,140 So 478 00:18:18,619 --> 00:18:20,779 Yeah. And and and and and, actually, there's 479 00:18:20,779 --> 00:18:22,400 starting to be blowback on that. 480 00:18:22,700 --> 00:18:24,059 I mean, if you go out on any 481 00:18:24,059 --> 00:18:26,720 social media site of whatever size or whatever, 482 00:18:27,194 --> 00:18:29,934 almost everybody who's a tech person will say, 483 00:18:30,474 --> 00:18:32,474 none of my locks are connected to the 484 00:18:32,474 --> 00:18:33,694 Internet. Thank you. 485 00:18:37,274 --> 00:18:38,494 And, you know, 486 00:18:39,034 --> 00:18:40,575 I think there's a fair deal of 487 00:18:40,954 --> 00:18:41,615 that resistance 488 00:18:42,630 --> 00:18:45,109 to opening up new markets now. We've we've 489 00:18:45,109 --> 00:18:46,869 just kind of done what we can in 490 00:18:46,869 --> 00:18:47,930 this tech mania, 491 00:18:48,710 --> 00:18:50,710 but but now it's kind of a more 492 00:18:50,710 --> 00:18:51,849 sober outlook. 493 00:18:52,150 --> 00:18:54,710 And that more sober outlook actually points to 494 00:18:54,710 --> 00:18:55,690 market saturation. 495 00:18:56,565 --> 00:18:57,065 And 496 00:18:57,365 --> 00:18:59,684 so v six is not, oh my god, 497 00:18:59,684 --> 00:19:01,365 there's this huge amount of v four that 498 00:19:01,365 --> 00:19:02,345 we've gotta cover. 499 00:19:04,964 --> 00:19:05,464 So 500 00:19:06,164 --> 00:19:07,224 what's the prospects? 501 00:19:08,619 --> 00:19:10,140 Yeah. So so that's why I asked about 502 00:19:10,140 --> 00:19:11,819 v six and the space it takes. Right? 503 00:19:11,819 --> 00:19:13,679 So if you take a slash 32, 504 00:19:14,299 --> 00:19:16,460 you could I don't know what how what 505 00:19:16,460 --> 00:19:18,299 is it? It's a slash one twenty eight. 506 00:19:18,299 --> 00:19:20,140 So even if you take the slash 32 507 00:19:20,140 --> 00:19:22,984 out, you know, you're looking at 4,200,000,000 508 00:19:23,125 --> 00:19:24,264 ish addresses 509 00:19:24,724 --> 00:19:27,204 in the slash 32 that are usable, even 510 00:19:27,204 --> 00:19:28,884 if you're doing slash 60 fours on all 511 00:19:28,884 --> 00:19:29,625 your hosts. 512 00:19:30,164 --> 00:19:30,904 So, like, 513 00:19:31,204 --> 00:19:33,204 you don't need to advertise a lot of 514 00:19:33,204 --> 00:19:35,524 slash 30 twos on the global on the 515 00:19:35,524 --> 00:19:36,024 DFZ 516 00:19:37,340 --> 00:19:39,580 to cover a lot of devices. Ah, the 517 00:19:39,580 --> 00:19:41,820 misery that is the misery that is v 518 00:19:41,820 --> 00:19:42,320 six. 519 00:19:42,859 --> 00:19:45,359 So if you're if you're a router vendor, 520 00:19:46,539 --> 00:19:48,320 then you can't assume 521 00:19:49,694 --> 00:19:52,674 that everyone routes to slash 48 or smaller. 522 00:19:53,375 --> 00:19:55,234 You can't make your internals 523 00:19:55,615 --> 00:19:56,994 48 bits wide. 524 00:19:57,694 --> 00:20:00,414 Your lookup tables are a 28 525 00:20:00,414 --> 00:20:02,974 bits wide. They're four times the size of 526 00:20:02,974 --> 00:20:05,580 v four. So the dear old routing table 527 00:20:05,580 --> 00:20:07,740 in v six, which is now kicking around 528 00:20:07,740 --> 00:20:08,700 200,000 529 00:20:08,700 --> 00:20:09,200 entries, 530 00:20:10,220 --> 00:20:11,359 is the equivalent 531 00:20:12,059 --> 00:20:13,259 of 800,000 532 00:20:13,259 --> 00:20:15,840 v four entries. It's the same size. 533 00:20:17,515 --> 00:20:20,075 And as I said, the sad fact is 534 00:20:20,075 --> 00:20:21,855 no one routes one twenty eights. 535 00:20:22,234 --> 00:20:24,555 No one routes 60 fours. We don't accept 536 00:20:24,555 --> 00:20:27,835 that. So most of that massive silicon out 537 00:20:27,835 --> 00:20:30,734 there sitting inside all these forwarding tables, 538 00:20:31,289 --> 00:20:33,710 in v four's in v six's case, 539 00:20:34,410 --> 00:20:36,269 two thirds is zero. 540 00:20:37,289 --> 00:20:37,789 Zero. 541 00:20:38,570 --> 00:20:40,750 You know? You're not actually using 542 00:20:41,049 --> 00:20:42,190 the bottom bits. 543 00:20:43,625 --> 00:20:45,884 Yep. And it it leads to the 544 00:20:46,424 --> 00:20:48,044 rather weird conclusion 545 00:20:49,144 --> 00:20:51,065 that V 6 is the same size as 546 00:20:51,065 --> 00:20:52,444 V 4, really. 547 00:20:54,744 --> 00:20:56,609 And and if we think, oh, well, we've 548 00:20:56,609 --> 00:20:59,009 filled up V 4, and we have, let's 549 00:20:59,009 --> 00:21:00,150 turn to V 6. 550 00:21:01,730 --> 00:21:02,630 But hang on. 551 00:21:03,329 --> 00:21:05,029 It's the same size, dude. 552 00:21:05,650 --> 00:21:06,789 It's no bigger. 553 00:21:07,970 --> 00:21:10,369 It doesn't give you huge amounts of space 554 00:21:10,369 --> 00:21:12,875 because we zeroed that out and don't let 555 00:21:12,875 --> 00:21:13,934 people route it. 556 00:21:15,434 --> 00:21:16,894 So we give homes 557 00:21:17,275 --> 00:21:18,654 a slash 48. 558 00:21:19,835 --> 00:21:22,494 Let me understand this. We give them 559 00:21:23,035 --> 00:21:26,414 how many bits of addressable space per home? 560 00:21:29,160 --> 00:21:31,500 80 bits. Yeah. I guess that's yeah. Yeah. 561 00:21:31,559 --> 00:21:33,820 We're, yeah, we're giving away slash 40 eights 562 00:21:34,039 --> 00:21:36,859 to most networks even at the enterprise level, 563 00:21:37,240 --> 00:21:38,539 not even to a campus. 564 00:21:39,174 --> 00:21:41,335 But, like, I just like, if I have 565 00:21:41,335 --> 00:21:43,575 a data center, I give the data center 566 00:21:43,575 --> 00:21:46,154 a slash, I don't know, 42 567 00:21:46,615 --> 00:21:48,695 or something like that or slash 40 or 568 00:21:48,695 --> 00:21:51,174 something. Right? Yeah. And I use slash 40 569 00:21:51,174 --> 00:21:52,154 eights for everything. 570 00:21:52,950 --> 00:21:55,509 My house. Slash 48. I don't have two 571 00:21:55,509 --> 00:21:57,289 to the eightieth bits of silicon. 572 00:21:57,909 --> 00:22:00,470 I can't imagine Yeah. A a world where 573 00:22:00,470 --> 00:22:02,789 even that would be possible. What is it? 574 00:22:02,789 --> 00:22:04,950 Awesome. My my my summary my summary of 575 00:22:04,950 --> 00:22:06,630 that is it's pretty easy to give away 576 00:22:06,630 --> 00:22:07,130 addresses. 577 00:22:07,595 --> 00:22:09,515 It's pretty hard to to and it's impossible 578 00:22:09,515 --> 00:22:10,575 to give away TCAM. 579 00:22:11,195 --> 00:22:11,695 Right? 580 00:22:12,154 --> 00:22:13,674 Like and I think that's I I don't 581 00:22:13,674 --> 00:22:15,755 know. I I find it funny. I would 582 00:22:15,755 --> 00:22:17,994 have assumed that people designing all these all 583 00:22:17,994 --> 00:22:19,615 the aggregation and all the rules, 584 00:22:20,369 --> 00:22:22,609 I find it funny that they seemingly were 585 00:22:22,609 --> 00:22:24,450 oblivious that there is a physical world in 586 00:22:24,450 --> 00:22:26,869 which this has to live. Like Nobody's 587 00:22:27,170 --> 00:22:27,910 in charge. 588 00:22:29,329 --> 00:22:31,589 Just remember that. Nobody's in charge. 589 00:22:32,130 --> 00:22:34,789 All of this works by mumblings in markets. 590 00:22:36,794 --> 00:22:39,534 And, you know, the address allocation policy 591 00:22:39,914 --> 00:22:42,474 was not driven by folk who make silicon 592 00:22:42,474 --> 00:22:44,015 devices that switch packets. 593 00:22:44,554 --> 00:22:46,875 Was not. Well, I'm not sure now that 594 00:22:46,875 --> 00:22:49,355 you say this. Maybe it was. Maybe it 595 00:22:49,355 --> 00:22:51,509 was maybe it was a clever marketing 596 00:22:51,809 --> 00:22:52,309 campaign. 597 00:22:53,410 --> 00:22:55,109 Cisco would be fine with all. 598 00:22:55,490 --> 00:22:55,990 Beautiful. 599 00:22:56,289 --> 00:22:58,289 Oh, I am. It was them. It was 600 00:22:58,289 --> 00:23:00,450 them that did this insanity. But, you know, 601 00:23:00,450 --> 00:23:02,769 the the issue was no one could kind 602 00:23:02,769 --> 00:23:05,714 of force the issue. And if one particular 603 00:23:05,775 --> 00:23:07,554 area goes, I've got pain, 604 00:23:08,015 --> 00:23:10,194 everyone else goes, get over it. 605 00:23:11,294 --> 00:23:13,875 And so we get these odd, just absolutely 606 00:23:14,095 --> 00:23:14,914 odd outcomes. 607 00:23:15,454 --> 00:23:17,315 Oh, we've got so many addresses. 608 00:23:17,830 --> 00:23:20,869 Let's rebuild v six in the precise image 609 00:23:20,869 --> 00:23:21,609 of v4. 610 00:23:22,470 --> 00:23:23,609 Precise image. 611 00:23:23,910 --> 00:23:25,930 So there's actually no incremental benefit. 612 00:23:27,910 --> 00:23:29,130 And then you wonder 613 00:23:29,750 --> 00:23:30,250 why 614 00:23:30,744 --> 00:23:32,825 the folk who are currently running v four, 615 00:23:32,825 --> 00:23:35,304 and it's around two thirds of users out 616 00:23:35,304 --> 00:23:35,804 there, 617 00:23:36,184 --> 00:23:38,505 don't have v six. You wonder why they 618 00:23:38,505 --> 00:23:40,284 don't feel any pressure to move. 619 00:23:41,224 --> 00:23:42,924 There's no new market. It works. 620 00:23:43,250 --> 00:23:45,730 Amazon just did this enormous amount of v 621 00:23:45,730 --> 00:23:47,910 four address space into the routing table. 622 00:23:49,570 --> 00:23:51,029 What's their v six plan? 623 00:23:51,330 --> 00:23:53,090 I don't know. It works. Their v four 624 00:23:53,090 --> 00:23:53,670 is is 625 00:23:54,850 --> 00:23:56,789 is still there and still going gangbusters. 626 00:23:57,974 --> 00:24:00,315 And so we're we're sitting in this sort 627 00:24:00,454 --> 00:24:01,994 of dual stack world, 628 00:24:02,535 --> 00:24:04,714 and we've hit this dual stack world 629 00:24:05,095 --> 00:24:07,575 at the point where market saturation and Moore's 630 00:24:07,575 --> 00:24:09,414 law coming to an end come to a 631 00:24:09,414 --> 00:24:09,914 nexus. 632 00:24:11,410 --> 00:24:13,009 And what I think we're going to see 633 00:24:13,009 --> 00:24:13,509 is 634 00:24:14,049 --> 00:24:16,549 this transition to a v six only network, 635 00:24:16,930 --> 00:24:19,590 we will continue to bleat on wood 636 00:24:20,529 --> 00:24:22,549 and mumble about, you know, in some halcyon 637 00:24:22,609 --> 00:24:24,289 day in the future, it'll be v six 638 00:24:24,289 --> 00:24:24,789 only, 639 00:24:25,544 --> 00:24:27,784 Perhaps not in my lifetime, all my kids, 640 00:24:27,784 --> 00:24:30,365 maybe my grandkids, you know, blah, blah, blah. 641 00:24:30,744 --> 00:24:32,044 But nothing will happen 642 00:24:32,664 --> 00:24:34,365 because there's no forcing function. 643 00:24:35,144 --> 00:24:38,365 Yeah. And even if there were, I worry, 644 00:24:39,139 --> 00:24:39,639 personally, 645 00:24:40,019 --> 00:24:40,579 because these 646 00:24:41,299 --> 00:24:42,740 so if you think about if you think 647 00:24:42,740 --> 00:24:45,380 about protocols like as having a WASP waste, 648 00:24:45,380 --> 00:24:48,259 right? Yeah. V four sits on top of 649 00:24:48,419 --> 00:24:50,419 or IP sits on top of all the 650 00:24:50,419 --> 00:24:50,919 strange 651 00:24:51,664 --> 00:24:52,804 physical layer stuff 652 00:24:53,184 --> 00:24:55,345 and makes all the strange physical layer stuff 653 00:24:55,345 --> 00:24:57,924 look the same, whether it's a LOHA, SONNET, 654 00:24:58,784 --> 00:24:59,284 token, 655 00:24:59,664 --> 00:25:01,525 token bus, token ring, 656 00:25:02,704 --> 00:25:03,204 all 657 00:25:03,505 --> 00:25:04,565 those all those wonderful 658 00:25:04,944 --> 00:25:05,845 frame relay, 659 00:25:06,390 --> 00:25:08,490 ATF, all those wonderful technologies. 660 00:25:09,109 --> 00:25:11,369 Oh, Ethernet too, by the way. But anyway 661 00:25:12,390 --> 00:25:12,789 but 662 00:25:13,190 --> 00:25:15,269 so so the so IP sits on top 663 00:25:15,269 --> 00:25:17,190 of all that. And then everything else sits 664 00:25:17,190 --> 00:25:18,869 on top of IP, whether it's quick or 665 00:25:18,869 --> 00:25:19,930 whatever it is. 666 00:25:20,744 --> 00:25:22,684 Switching IP for real 667 00:25:24,265 --> 00:25:24,924 is like, 668 00:25:25,384 --> 00:25:27,144 I don't know, it's like it's like pulling 669 00:25:27,144 --> 00:25:27,964 out the centerpiece 670 00:25:28,265 --> 00:25:31,005 in a two foot high Jenga stack 671 00:25:31,865 --> 00:25:33,724 and putting the piece back in 672 00:25:34,079 --> 00:25:36,480 without it falling over, not just pulling it 673 00:25:36,480 --> 00:25:38,559 out. You gotta put a new one in 674 00:25:38,559 --> 00:25:40,740 and get the old one out. Like, 675 00:25:41,119 --> 00:25:41,619 what? 676 00:25:43,200 --> 00:25:45,279 Right. It's not that we're addicted to v 677 00:25:45,279 --> 00:25:47,220 four. We're glued into it. 678 00:25:47,875 --> 00:25:50,194 Yeah. It's it's gonna be hard. It's gonna 679 00:25:50,194 --> 00:25:51,595 be hard. It is hard, and that's what 680 00:25:51,714 --> 00:25:54,355 why we're willing to pay. And and the 681 00:25:54,355 --> 00:25:55,335 transfer market 682 00:25:55,954 --> 00:25:56,934 is now quoting 683 00:25:57,714 --> 00:25:58,214 around, 684 00:25:59,899 --> 00:26:02,220 on average between 30 and $35 685 00:26:02,220 --> 00:26:03,919 an IP address for v four. 686 00:26:04,700 --> 00:26:06,159 And everyone's paying it. 687 00:26:06,700 --> 00:26:07,679 They're paying it. 688 00:26:08,059 --> 00:26:08,880 And so 689 00:26:09,339 --> 00:26:10,159 this kind 690 00:26:11,019 --> 00:26:13,759 of fixation on v four is not accidental. 691 00:26:13,980 --> 00:26:16,575 It's we've built it this way, and it's 692 00:26:16,634 --> 00:26:17,934 really, really hard 693 00:26:18,315 --> 00:26:20,714 to actually back out of that and look 694 00:26:20,714 --> 00:26:21,994 at it again and go, well, I can 695 00:26:21,994 --> 00:26:23,375 replace that with v six. 696 00:26:23,835 --> 00:26:27,535 Only green fields, only green fields actually have 697 00:26:27,740 --> 00:26:30,059 the ability to go right from scratch. It's 698 00:26:30,059 --> 00:26:31,039 v six only. 699 00:26:31,500 --> 00:26:33,200 But even they have a problem. 700 00:26:33,900 --> 00:26:36,380 Like, the Indians have done it. By and 701 00:26:36,380 --> 00:26:38,720 large, the subcontinent is v six only. 702 00:26:39,420 --> 00:26:40,779 And they go, well, I want to use 703 00:26:40,779 --> 00:26:41,279 Amazon. 704 00:26:42,744 --> 00:26:44,345 I want to use Slack. I want to 705 00:26:44,345 --> 00:26:46,105 use all these apps. And they go, oh, 706 00:26:46,105 --> 00:26:46,605 sorry. 707 00:26:47,144 --> 00:26:49,085 We just don't have the V 6 capacity. 708 00:26:49,144 --> 00:26:50,444 You're going to have to somehow 709 00:26:50,825 --> 00:26:53,144 map back into V 4 to get access 710 00:26:53,144 --> 00:26:54,044 to these services. 711 00:26:54,710 --> 00:26:57,450 There's no one out there with a retail 712 00:26:57,990 --> 00:26:58,490 offering 713 00:26:58,950 --> 00:27:00,730 that's doing it in V 6 only. 714 00:27:01,910 --> 00:27:03,509 And that's the kind of the big the 715 00:27:03,509 --> 00:27:04,329 the big problem, 716 00:27:04,630 --> 00:27:06,309 that all of us have to sort of 717 00:27:06,309 --> 00:27:08,970 keep pools of V 4 addresses lying around 718 00:27:09,394 --> 00:27:10,994 in order to make sure that we have 719 00:27:10,994 --> 00:27:12,134 universal connectivity. 720 00:27:12,835 --> 00:27:14,994 If you just used V 6, it isn't 721 00:27:14,994 --> 00:27:16,214 gonna work that well. 722 00:27:17,234 --> 00:27:20,194 Oops. Yeah. And what's and what's rather humorous 723 00:27:20,194 --> 00:27:22,054 in some of these things about it is 724 00:27:22,319 --> 00:27:24,399 some of these networks are actually v six 725 00:27:24,399 --> 00:27:25,220 only internally. 726 00:27:26,880 --> 00:27:28,640 Alternate. And they're tunneling v four. Yeah. They're 727 00:27:28,640 --> 00:27:31,599 tunneling v four on top. Externally, they're v 728 00:27:31,599 --> 00:27:34,259 four. But internally, they're v six. 729 00:27:35,039 --> 00:27:35,779 And, like, 730 00:27:36,480 --> 00:27:37,220 that's how 731 00:27:37,904 --> 00:27:40,325 that's how universal v four is 732 00:27:41,505 --> 00:27:43,684 as an as a way of of accessing 733 00:27:43,744 --> 00:27:44,244 things. 734 00:27:44,865 --> 00:27:47,744 So over in the RIR land where I 735 00:27:47,744 --> 00:27:50,325 work, where we hand out v six addresses, 736 00:27:51,424 --> 00:27:53,125 the last two years, 737 00:27:53,930 --> 00:27:56,730 twenty twenty three, twenty twenty four, would actually 738 00:27:56,730 --> 00:27:57,230 have 739 00:27:57,609 --> 00:27:58,509 the smallest 740 00:27:58,890 --> 00:28:00,910 allocations of v six addresses 741 00:28:01,930 --> 00:28:03,309 for the last ten years, 742 00:28:04,170 --> 00:28:04,670 except 743 00:28:05,775 --> 00:28:06,275 except 744 00:28:07,215 --> 00:28:08,275 two allocations 745 00:28:09,855 --> 00:28:11,795 completely tipped it the other way. 746 00:28:12,654 --> 00:28:14,115 A mid level bank, 747 00:28:14,815 --> 00:28:16,515 in The US last year 748 00:28:16,894 --> 00:28:18,755 had I think it was a slash 16 749 00:28:18,894 --> 00:28:19,875 in v six, 750 00:28:20,255 --> 00:28:22,369 an awesomely massive allocation. 751 00:28:22,670 --> 00:28:24,190 It's as if they wanted to put a 752 00:28:24,190 --> 00:28:26,590 v six address on every credit card they 753 00:28:26,590 --> 00:28:28,130 issued and more. 754 00:28:30,029 --> 00:28:31,490 That single allocation 755 00:28:32,190 --> 00:28:33,570 was kind of double 756 00:28:33,894 --> 00:28:36,454 everything else we'd done in the previous two 757 00:28:36,454 --> 00:28:36,954 years. 758 00:28:37,255 --> 00:28:37,755 Massive. 759 00:28:38,615 --> 00:28:39,674 And then this year, 760 00:28:40,375 --> 00:28:41,434 over in APNIC, 761 00:28:42,454 --> 00:28:44,375 it's common in fact, it's it's public. All 762 00:28:44,375 --> 00:28:46,295 allocations are public knowledge, so I'm not telling 763 00:28:46,295 --> 00:28:49,740 any secrets. Huawei came out with, again, around, 764 00:28:49,740 --> 00:28:51,500 I think, it's a slash a t. It's 765 00:28:51,500 --> 00:28:52,000 huge. 766 00:28:53,259 --> 00:28:54,880 And it's those two allocations 767 00:28:55,259 --> 00:28:55,759 alone 768 00:28:56,779 --> 00:28:59,580 that have changed the v six picture in 769 00:28:59,580 --> 00:29:02,220 aggregate from, oh my god. No one's interested 770 00:29:02,220 --> 00:29:04,105 in v six to we've just had the 771 00:29:04,105 --> 00:29:06,924 two largest years of address allocations ever, dude, 772 00:29:07,544 --> 00:29:08,845 thanks to just two allocations, 773 00:29:09,464 --> 00:29:10,285 just two. 774 00:29:10,664 --> 00:29:13,005 And, again, this is really hard to explain. 775 00:29:13,544 --> 00:29:15,804 A small number of folk are doing awesomely 776 00:29:15,865 --> 00:29:17,384 big bets on v six, and the rest 777 00:29:17,384 --> 00:29:18,444 of the world's going, 778 00:29:18,929 --> 00:29:20,470 nah, I haven't got a problem. 779 00:29:22,129 --> 00:29:24,929 Not gonna change. Yeah. Now, I do think 780 00:29:24,929 --> 00:29:25,429 SRv6, 781 00:29:25,809 --> 00:29:27,190 if people deploy it, 782 00:29:27,490 --> 00:29:29,109 will cause some v six 783 00:29:29,970 --> 00:29:30,470 allocations, 784 00:29:31,505 --> 00:29:32,884 but they won't be publicly 785 00:29:33,424 --> 00:29:33,924 advertised. 786 00:29:34,305 --> 00:29:35,285 And that's another 787 00:29:35,664 --> 00:29:38,945 big question. Right? There'll be allocations for internal 788 00:29:38,945 --> 00:29:39,765 use only. 789 00:29:40,384 --> 00:29:42,085 And we've already got, 790 00:29:42,785 --> 00:29:45,345 unique local addresses. We've already got the equivalent 791 00:29:45,345 --> 00:29:48,109 of private address space. So to what extent 792 00:29:48,170 --> 00:29:50,569 folk are willing to expose their plans, like 793 00:29:50,569 --> 00:29:51,690 Huawei and this, 794 00:29:52,410 --> 00:29:54,809 rather odd American bank and do a public 795 00:29:54,809 --> 00:29:55,710 address allocation 796 00:29:56,329 --> 00:29:58,490 versus just simply going, oh, I'm just using 797 00:29:58,490 --> 00:29:59,390 private addresses. 798 00:29:59,690 --> 00:30:02,089 Not gonna tell anyone. It's all secret. Well, 799 00:30:02,089 --> 00:30:04,585 yeah. Three f f f has changed that 800 00:30:04,585 --> 00:30:06,505 picture a little bit. Like, the u EULA 801 00:30:06,505 --> 00:30:09,144 addresses are pretty useless for for v for 802 00:30:09,144 --> 00:30:11,884 SRV six, but the three f f addresses 803 00:30:12,025 --> 00:30:12,765 might be 804 00:30:13,305 --> 00:30:14,984 might not actually be something you could use. 805 00:30:14,984 --> 00:30:16,744 It's a big pool, and and who knows 806 00:30:16,904 --> 00:30:18,789 Yeah. Is what I'm saying. But, you know, 807 00:30:18,789 --> 00:30:20,390 part of the issue around with v six 808 00:30:20,390 --> 00:30:20,890 is 809 00:30:22,950 --> 00:30:24,250 part of market saturation, 810 00:30:24,549 --> 00:30:26,950 part of the difference in the change from 811 00:30:26,950 --> 00:30:30,490 network centric networking to cloud centric networking 812 00:30:31,109 --> 00:30:32,650 actually means that the 813 00:30:33,315 --> 00:30:35,414 target of routing and addressing 814 00:30:35,954 --> 00:30:36,694 is lower. 815 00:30:37,634 --> 00:30:39,974 We deal in a namespace, in an application 816 00:30:40,034 --> 00:30:40,534 space. 817 00:30:41,075 --> 00:30:43,315 There's less and less emphasis on what happens 818 00:30:43,315 --> 00:30:44,534 at the network level. 819 00:30:45,089 --> 00:30:45,589 You 820 00:30:46,609 --> 00:30:49,089 know, there's this theory out there that once 821 00:30:49,089 --> 00:30:50,789 the cloud folk have their way, 822 00:30:51,089 --> 00:30:53,029 the evil people that they are, 823 00:30:53,329 --> 00:30:54,630 there will be no routing. 824 00:30:56,210 --> 00:30:58,390 The clouds will have their points of presence 825 00:30:59,014 --> 00:30:59,835 right beside 826 00:31:00,294 --> 00:31:02,954 every access network in every data center. 827 00:31:03,494 --> 00:31:05,734 And so there's no point to move packets 828 00:31:05,734 --> 00:31:08,134 around the world. That's all done in private 829 00:31:08,134 --> 00:31:10,794 networks inside various people's clouds. 830 00:31:11,174 --> 00:31:14,369 It's not you. It's not me. Right. And 831 00:31:14,369 --> 00:31:16,130 and then it's kinda And and then it 832 00:31:16,130 --> 00:31:17,190 becomes all proprietary, 833 00:31:17,970 --> 00:31:20,529 using whatever they wanna use, and all the 834 00:31:20,529 --> 00:31:22,130 open standards go the way of the wind 835 00:31:22,130 --> 00:31:23,910 because why do you need open standards? 836 00:31:24,450 --> 00:31:26,690 Open standards are just an interface into a 837 00:31:26,690 --> 00:31:27,190 cloud. 838 00:31:28,065 --> 00:31:30,325 Exactly. And and we're seeing inexorably 839 00:31:30,704 --> 00:31:32,544 as a market and, again, this is all 840 00:31:32,544 --> 00:31:33,524 just not planned. 841 00:31:33,825 --> 00:31:35,825 This is just everyone trying to optimize, you 842 00:31:35,825 --> 00:31:37,984 know, where they wanna go. As a market, 843 00:31:37,984 --> 00:31:39,365 that's where we're heading. 844 00:31:40,690 --> 00:31:43,490 And that's one of the most fascinating change 845 00:31:43,490 --> 00:31:45,430 in the last, I'll call it a decade. 846 00:31:46,289 --> 00:31:48,950 It's much more fundamental than v six. 847 00:31:49,330 --> 00:31:50,710 V six is a 1980s 848 00:31:51,090 --> 00:31:53,029 architecture the same as v four. 849 00:31:53,410 --> 00:31:56,345 Changing the address space, oddly enough, painful as 850 00:31:56,345 --> 00:31:58,684 it has been, was not a fundamental change. 851 00:31:59,464 --> 00:32:01,244 But this whole issue of 852 00:32:01,784 --> 00:32:02,845 making names 853 00:32:03,304 --> 00:32:05,625 much more important and moving into the cloud 854 00:32:05,625 --> 00:32:06,125 space 855 00:32:06,664 --> 00:32:09,865 is absolutely basic to the way we look 856 00:32:09,865 --> 00:32:10,605 at networking. 857 00:32:11,119 --> 00:32:12,340 Do we need standards? 858 00:32:12,880 --> 00:32:15,059 Oh, in a cloud world, as you said, 859 00:32:15,440 --> 00:32:16,580 maybe, maybe not. 860 00:32:17,440 --> 00:32:19,200 I wanna do SR v six. Well, you 861 00:32:19,200 --> 00:32:21,220 go do it, dude. I don't need to. 862 00:32:22,320 --> 00:32:23,575 Oh. Oh, 863 00:32:23,875 --> 00:32:25,954 I wanna run a new variant of BGP. 864 00:32:25,954 --> 00:32:28,355 Well, you know, knock yourself out. I'm not 865 00:32:28,355 --> 00:32:30,214 a fan. Yeah. Those open source implementations 866 00:32:30,515 --> 00:32:31,174 do it. 867 00:32:31,954 --> 00:32:33,394 Exactly. But the thing is, you can do 868 00:32:33,394 --> 00:32:34,755 it in your little neck of the woods, 869 00:32:34,755 --> 00:32:36,115 and I can do a different thing in 870 00:32:36,115 --> 00:32:38,929 mine because we don't have to exchange packets 871 00:32:38,929 --> 00:32:41,409 at the network level. We exchange, if you 872 00:32:41,409 --> 00:32:42,470 will, application 873 00:32:42,849 --> 00:32:44,629 packets at the name level. 874 00:32:45,809 --> 00:32:47,589 Yep. Exactly. Yep. 875 00:32:48,129 --> 00:32:49,970 And so it's it's an odd world that 876 00:32:49,970 --> 00:32:52,464 we're sitting in that we're being driven into, 877 00:32:52,464 --> 00:32:53,204 not deliberately, 878 00:32:53,744 --> 00:32:57,045 but just millions of small self optimizing decisions 879 00:32:57,664 --> 00:33:00,884 produce this momentum that actually walks away from 880 00:33:01,265 --> 00:33:03,585 most of the common routing and addressing fabric 881 00:33:03,585 --> 00:33:04,325 of the internet. 882 00:33:06,410 --> 00:33:09,210 Wow. We got there in forty minutes to 883 00:33:09,210 --> 00:33:12,170 such a massive conceptual sort of result. This 884 00:33:12,170 --> 00:33:14,329 is kind of, oh my god. You mean 885 00:33:14,329 --> 00:33:15,789 everything I learned is irrelevant? 886 00:33:18,970 --> 00:33:19,789 Yes. Alright. 887 00:33:20,194 --> 00:33:21,255 Let's move on. 888 00:33:21,875 --> 00:33:22,535 Yeah. Right. 889 00:33:23,075 --> 00:33:25,015 Didn't happen. That makes us uncomfortable. 890 00:33:28,035 --> 00:33:30,615 Yeah. I think it's worth looking at. Yeah. 891 00:33:30,674 --> 00:33:32,755 Yeah. Yeah. So maybe we should get you 892 00:33:32,755 --> 00:33:34,819 back on to talk about this again. I 893 00:33:34,819 --> 00:33:36,339 don't know. I mean, we always like to 894 00:33:36,339 --> 00:33:38,660 have you on anyway, Jeff. Look. I I 895 00:33:38,660 --> 00:33:40,119 do my best in all this, 896 00:33:40,819 --> 00:33:42,500 and it's kind of trying to match the 897 00:33:42,500 --> 00:33:43,000 minutia 898 00:33:43,539 --> 00:33:46,500 of what's happening with individual addresses, with routing 899 00:33:46,500 --> 00:33:48,704 tables, with the amount of churn out there, 900 00:33:48,865 --> 00:33:51,184 what's happening in terms of moving packets around 901 00:33:51,184 --> 00:33:53,684 the world, but at the same time asking, 902 00:33:54,065 --> 00:33:55,984 why are these packets moving around the world 903 00:33:55,984 --> 00:33:57,845 and literally are they anymore? 904 00:33:59,024 --> 00:34:01,505 You know, I've spoken myself and turned a 905 00:34:01,505 --> 00:34:03,765 lot more attention to the domain name system. 906 00:34:04,210 --> 00:34:06,210 I actually feel that these days, there's so 907 00:34:06,210 --> 00:34:07,029 many NATs, 908 00:34:07,330 --> 00:34:09,670 the whole bifurcation in address space. 909 00:34:10,449 --> 00:34:12,849 Does being part of one address space matter 910 00:34:12,849 --> 00:34:14,849 anymore? I look around my home and I 911 00:34:14,849 --> 00:34:17,429 haven't got an IP address that's publicly unique. 912 00:34:17,964 --> 00:34:19,025 I'm behind NATs. 913 00:34:19,565 --> 00:34:21,244 The bits of v six that I have 914 00:34:21,244 --> 00:34:23,565 are so heavily they look like a NAT 915 00:34:23,565 --> 00:34:25,105 even though they're not because, 916 00:34:25,484 --> 00:34:27,885 you know, NATs are are secure, aren't they? 917 00:34:27,885 --> 00:34:29,724 And so v six is built in NAT's 918 00:34:29,724 --> 00:34:30,224 image. 919 00:34:30,539 --> 00:34:33,260 I've got nothing that's accessible inside my house 920 00:34:33,260 --> 00:34:33,760 anymore, 921 00:34:34,300 --> 00:34:36,059 and I'm okay with that, and so is 922 00:34:36,059 --> 00:34:36,880 everyone else. 923 00:34:37,980 --> 00:34:40,219 It's In fact, it's good. It's good. This 924 00:34:40,219 --> 00:34:42,219 is an argument that I had very early 925 00:34:42,219 --> 00:34:44,164 on when they were talking about NATs and 926 00:34:44,164 --> 00:34:45,864 how they weren't a security mechanism. 927 00:34:46,405 --> 00:34:49,525 Okay. I get that they're not purely a 928 00:34:49,525 --> 00:34:52,244 security mechanism. I understand that. Right? Yeah. They're 929 00:34:52,244 --> 00:34:55,704 not in security terms. They're not a true 930 00:34:55,844 --> 00:34:57,465 security mechanism. Right? 931 00:34:58,250 --> 00:35:00,269 But let's be honest, they're fail open. 932 00:35:01,849 --> 00:35:02,349 And 933 00:35:02,650 --> 00:35:04,750 not only they not only fail open, 934 00:35:05,130 --> 00:35:06,989 but they actually they do obfuscate. 935 00:35:07,369 --> 00:35:08,969 And as I said on my mailing list 936 00:35:08,969 --> 00:35:11,789 once, do you not put curtains 937 00:35:12,664 --> 00:35:13,644 in your windows? 938 00:35:14,425 --> 00:35:17,864 Everywhere except for that. Everywhere except for the 939 00:35:17,864 --> 00:35:18,684 good touch. 940 00:35:19,704 --> 00:35:21,085 You know, it's like 941 00:35:21,784 --> 00:35:23,244 Yeah. Yeah. Obfuscation 942 00:35:23,704 --> 00:35:24,525 is important. 943 00:35:24,829 --> 00:35:26,750 Okay? It is a security. It is a 944 00:35:26,750 --> 00:35:29,389 component of a larger security thing. So don't 945 00:35:29,389 --> 00:35:31,230 tell me how useless it is because it 946 00:35:31,230 --> 00:35:32,130 does matter. 947 00:35:32,750 --> 00:35:34,590 It all matters. Where I'm getting to is 948 00:35:34,590 --> 00:35:37,090 this kind of bifurcation of big pictures, detailed 949 00:35:37,150 --> 00:35:37,650 picture 950 00:35:37,965 --> 00:35:40,445 on the Internet. Yeah. And the details point 951 00:35:40,445 --> 00:35:42,364 you down one area of, oh, yeah. The 952 00:35:42,364 --> 00:35:44,525 routing table's still growing. Oh, we're still handing 953 00:35:44,525 --> 00:35:46,545 out addresses. Oh, it's still important. 954 00:35:47,244 --> 00:35:48,945 The big picture kinda says, 955 00:35:49,405 --> 00:35:51,664 were there, dude? Not not so much. 956 00:35:52,320 --> 00:35:53,219 Not so much. 957 00:35:53,840 --> 00:35:55,219 Names are actually critical. 958 00:35:55,840 --> 00:35:57,700 Domain names define the Internet. 959 00:35:58,400 --> 00:35:59,219 And and 960 00:35:59,519 --> 00:36:01,680 this is kind of the underlying change that's 961 00:36:01,680 --> 00:36:03,380 happened. And and the reason why 962 00:36:04,079 --> 00:36:05,460 is actually really simple. 963 00:36:05,904 --> 00:36:09,125 Once you start putting supercomputers on everyone's wrist, 964 00:36:10,384 --> 00:36:13,125 that's filled with software and data. They're 965 00:36:13,744 --> 00:36:14,964 enormously capable. 966 00:36:15,664 --> 00:36:17,585 They don't need to pay the network a 967 00:36:17,585 --> 00:36:19,039 tithe to get 968 00:36:19,500 --> 00:36:21,660 servers. They just extract the servers the way 969 00:36:21,660 --> 00:36:22,719 they want it because 970 00:36:23,820 --> 00:36:25,820 it's a supercomputer, dude. It can do what 971 00:36:25,820 --> 00:36:27,119 it wants on my wrist, 972 00:36:27,820 --> 00:36:29,980 you know, in my phone, on my desk, 973 00:36:29,980 --> 00:36:30,480 wherever. 974 00:36:30,984 --> 00:36:33,945 And and that change of pushing computing out 975 00:36:33,945 --> 00:36:36,264 to the edge has pushed value out to 976 00:36:36,264 --> 00:36:38,344 the edge and taken value out of the 977 00:36:38,344 --> 00:36:39,324 network middle. 978 00:36:40,744 --> 00:36:43,324 We're in a networking podcast, aren't we? 979 00:36:43,860 --> 00:36:45,940 Yeah. Yeah. Also, I think that's a small. 980 00:36:45,940 --> 00:36:47,240 Sorry about that. 981 00:36:50,099 --> 00:36:50,599 Yeah. 982 00:36:51,059 --> 00:36:52,820 Okay. We should wrap up and get you 983 00:36:52,820 --> 00:36:54,340 back on again, Jeff. We should try to 984 00:36:54,340 --> 00:36:56,340 get you on more often along with other 985 00:36:56,340 --> 00:36:56,980 people. But, 986 00:36:57,539 --> 00:36:58,519 cool. Alright. 987 00:36:59,574 --> 00:37:00,855 It's a good I guess I'll start with 988 00:37:01,414 --> 00:37:03,735 yeah. Where where do people find you? Poteroo. 989 00:37:03,735 --> 00:37:06,295 Right? C O H At Poteroo Dot Net 990 00:37:06,295 --> 00:37:07,974 will find me, or you can just look 991 00:37:07,974 --> 00:37:09,815 on the web page for all awesome stuff 992 00:37:09,815 --> 00:37:10,635 about BGB 993 00:37:11,015 --> 00:37:15,550 and my, you know, insane ramblings on www.potaroo.net. 994 00:37:16,489 --> 00:37:16,989 Cool. 995 00:37:17,289 --> 00:37:19,789 Awesome. And I assume you're on LinkedIn and 996 00:37:19,849 --> 00:37:20,750 blah dee da. 997 00:37:21,369 --> 00:37:23,710 Less so these days. I'll be very fashionable. 998 00:37:28,724 --> 00:37:31,445 Alright. Cool. And, Tom, where can people reach 999 00:37:31,445 --> 00:37:33,045 you? LinkedIn. I'll just go ahead and say 1000 00:37:35,765 --> 00:37:37,204 I even talked over, so you have to 1001 00:37:37,204 --> 00:37:38,010 say it again. 1002 00:37:38,329 --> 00:37:40,429 LinkedIn. There I put it down for 1003 00:37:42,170 --> 00:37:42,829 you. Thanks. 1004 00:37:43,289 --> 00:37:46,250 Alright. Well, thanks, Jeff, for coming on. And 1005 00:37:46,250 --> 00:37:48,650 to all the folks out there listening, we 1006 00:37:48,650 --> 00:37:50,889 know that we live in an attention driven 1007 00:37:50,889 --> 00:37:52,969 economy, so your time is important to us, 1008 00:37:52,969 --> 00:37:54,554 and we're thankful we have an audience. 1009 00:37:55,015 --> 00:37:57,494 And thanks for listening to this episode of 1010 00:37:57,494 --> 00:37:59,094 The Hedge all the way to the very 1011 00:37:59,094 --> 00:38:00,534 bitter end. We're gonna have to talk about 1012 00:38:00,534 --> 00:38:03,414 what that term means eventually, but all the 1013 00:38:03,414 --> 00:38:04,934 way to the bitter end, and we will 1014 00:38:04,934 --> 00:38:06,074 catch you next time.