Node.js Performance, Kubernetes, and Why “Fast” Isn’t Always Fast - JSJ 702의 전체 녹취록

00:02:12.439 to a nightmare. 42 00:02:13.159 --> 00:02:15.360 Well, all I can say is will probably be in 43 00:02:15.400 --> 00:02:19.439 Italy again because we love Italy, and next time hopefully 44 00:02:19.479 --> 00:02:21.319 we'll be able to actually meet. 45 00:02:23.800 --> 00:02:26.080 Yeah, I would love to go back. We had a 46 00:02:26.120 --> 00:02:28.879 foreign exchange student who also lives in Oncona and so 47 00:02:29.240 --> 00:02:32.879 we'd like to see her and her baby. So anyway, 48 00:02:33.879 --> 00:02:36.680 what are we talking about that's not travel tech or 49 00:02:37.280 --> 00:02:38.479 Steve needs a job. 50 00:02:39.599 --> 00:02:42.280 Today we're talking a little bit about the latest thing 51 00:02:42.599 --> 00:02:48.360 in platformatic my company, and not j yes, I think 52 00:02:49.719 --> 00:02:53.080 maybe with some ais pist in it, because you know, 53 00:02:53.120 --> 00:02:54.199 everything is AI today. 54 00:02:54.319 --> 00:02:56.919 Yeah, a lot of our listeners may not be familiar 55 00:02:56.919 --> 00:03:00.039 with Platformatic, but I think that if you're doing a 56 00:03:00.199 --> 00:03:03.000 think with notes, there's a good chance that you're probably 57 00:03:03.080 --> 00:03:05.240 using at least one thing from Platformatic. 58 00:03:05.479 --> 00:03:08.960 Like it's absolutely at this point, I think you're using 59 00:03:09.000 --> 00:03:13.800 something from from Platformatic close to one hundred percent. Okay, 60 00:03:14.000 --> 00:03:19.879 it's as I say, last year, I touched forty two 61 00:03:20.240 --> 00:03:24.599 a billion downloads on NPM some some some impossible numbers 62 00:03:24.639 --> 00:03:24.919 like that. 63 00:03:25.599 --> 00:03:27.520 So this is the state. 64 00:03:28.960 --> 00:03:31.800 So like every man, woman and child on this planet 65 00:03:31.919 --> 00:03:36.000 is downloading stuff on Platformatic like ten times a year. 66 00:03:36.039 --> 00:03:36.800 I have no idea. 67 00:03:36.879 --> 00:03:40.960 Yeah, at this point, it's it's it's it's it's everywhere. 68 00:03:41.039 --> 00:03:44.879 As I say, I am in everybody Dependency three. 69 00:03:45.360 --> 00:03:47.280 So I am the Nebraska man. I'm joking. 70 00:03:51.280 --> 00:03:53.240 Okay, I thought it was just your mom over there 71 00:03:53.360 --> 00:03:55.280 NPM installed Platformatic stuff. 72 00:03:55.479 --> 00:03:57.639 Look, that can be possible. You know. 73 00:03:58.000 --> 00:04:01.280 I fixed that laptop for for christ So that was 74 00:04:01.280 --> 00:04:01.680 the gift. 75 00:04:02.280 --> 00:04:02.800 There you go. 76 00:04:04.039 --> 00:04:06.039 So it's the one time over the over the year 77 00:04:06.120 --> 00:04:10.680 that somebody can ask me to fix a computer over Christmas. 78 00:04:11.680 --> 00:04:12.400 The tech support. 79 00:04:13.680 --> 00:04:17.680 It's it's it's it's something that everybody does. So so 80 00:04:17.800 --> 00:04:22.639 true talking about things we have since the

last time 81 00:04:22.759 --> 00:04:27.519 was year so what we have released a few benchmarks 82 00:04:27.560 --> 00:04:34.319 okay that I think are pretty relevant for everybody to 83 00:04:35.399 --> 00:04:38.399 look at. So let me just get them and so 84 00:04:38.439 --> 00:04:43.120 that we can discuss them. I'm going to pass them. 85 00:04:43.040 --> 00:04:46.199 While you do. I was. I want to emphasize that 86 00:04:46.240 --> 00:04:51.199 while I was obviously somewhat kidding before, it is absolutely 87 00:04:51.240 --> 00:04:53.759 true that I kind of considue you to be sort 88 00:04:53.759 --> 00:04:57.360 of mister no JS, especially now that Ryan Dalla has 89 00:04:57.439 --> 00:04:58.920 moved on to other things. 90 00:04:59.800 --> 00:05:05.519 H It's well, I have been since this summer. I've 91 00:05:05.560 --> 00:05:09.560 been nominated the chair of the no JS project. So wow, 92 00:05:10.120 --> 00:05:13.240 I am effectively running the thing. 93 00:05:14.720 --> 00:05:17.560 Wow. I didn't even know that. Wow, that's pretty. 94 00:05:17.439 --> 00:05:23.399 It's an honor. Let's put it this way. It's an honor. 95 00:05:23.560 --> 00:05:26.439 Yeah, we talked about it slightly before we started the 96 00:05:26.480 --> 00:05:30.959 show officially that I'm I'm you know, Node is not new, 97 00:05:31.399 --> 00:05:35.560 obviously it's been around for a while, but I'm I'm 98 00:05:35.639 --> 00:05:41.199 seeing Node usage actually increasing. A lot of organizations that 99 00:05:41.240 --> 00:05:44.360 have previously been using other technologies for the back end 100 00:05:45.040 --> 00:05:48.879 are adopting Node as their go to technology for back 101 00:05:49.000 --> 00:05:52.279 end stuff. I don't know if it's because of the 102 00:05:52.399 --> 00:05:57.480 Lambda functions and stuff like that, or it's because you know, 103 00:05:57.600 --> 00:06:02.920 all the meta framework out there, Next JS and ten 104 00:06:03.000 --> 00:06:06.319 stack and whatnot. I don't know. Maybe it's like everything together, 105 00:06:06.879 --> 00:06:11.560 but I'm definitely seeing no JS technology usage pickup a 106 00:06:11.560 --> 00:06:15.040 lot of companies. Like I was speaking at this conference 107 00:06:15.079 --> 00:06:18.839 in Israel, and like every company that I was talking 108 00:06:18.879 --> 00:06:22.240 to was doing something with no JS on the back end. 109 00:06:23.240 --> 00:06:26.800 Yeah, it's everywhere, Okay, right now, if you want to 110 00:06:26.800 --> 00:06:28.879 build anything for the Web, you're using Note. 111 00:06:29.399 --> 00:06:32.959 If you're not using Note, you are probably. 112 00:06:32.560 --> 00:06:36.839 On a minority, and it's it might be good to 113 00:06:36.839 --> 00:06:40.319 be in a minority. But the mainstream stack right now 114 00:06:41.160 --> 00:06:45.480 is something React based plus no JS on the server. 115 00:06:48.079 --> 00:06:51.560 And I'm in the minority me too, I know. 116 00:06:52.000

--> 00:06:58.199 But you're probably using View. I've heard okay, Laravel Lvl, I. 117 00:06:58.160 --> 00:07:00.720 Would I would wonder, Okay, this is a how do 118 00:07:00.759 --> 00:07:03.800 you do server side rendering with view on Laavo. 119 00:07:04.879 --> 00:07:05.600 If you use. 120 00:07:05.879 --> 00:07:09.319 Inertia JS, which is something I'm a big fan of 121 00:07:09.360 --> 00:07:13.160 and I've talked about ad nauseum probably on here. Inertia 122 00:07:13.199 --> 00:07:15.680 has a service side rending capability. If you're using View 123 00:07:15.720 --> 00:07:18.560 with something like nuxt, it also has server side rending capability. 124 00:07:18.759 --> 00:07:21.600 Yeah, but then it's node right correct. 125 00:07:21.800 --> 00:07:24.399 Okay, point well, you asked how to do with view, 126 00:07:24.519 --> 00:07:25.600 so that's how No. 127 00:07:25.600 --> 00:07:27.879 No, But it's say with knox is node. 128 00:07:28.079 --> 00:07:28.240 Yes. 129 00:07:29.480 --> 00:07:33.000 Is the inertia not no, not JavaScript running somewhere. 130 00:07:33.240 --> 00:07:37.000 Inertia is real quick. Is basically a glue layer between 131 00:07:37.000 --> 00:07:38.399 your front and your back end that allows you to 132 00:07:38.399 --> 00:07:41.959 plug and play. So you could use View with Ruby, 133 00:07:42.000 --> 00:07:44.959 with Laravel, with Node, you could use you So you 134 00:07:44.959 --> 00:07:48.160 could use viewsvelt react Angular on the front end, and 135 00:07:48.240 --> 00:07:53.560 you could use laravl Node, Ruby on the back end 136 00:07:54.160 --> 00:07:59.199 in any combination. So it's just it's crazy fast. I 137 00:07:59.240 --> 00:08:01.879 have an application that I've built and maintained for a while, 138 00:08:01.879 --> 00:08:04.600 and it's just amazing how fast it is. It basically 139 00:08:04.720 --> 00:08:09.519 sort of hijacks the post request and the browser doesn't 140 00:08:09.519 --> 00:08:11.920 do a full reload. You passes some headers that tell it, 141 00:08:11.920 --> 00:08:14.639 hey don't do a full Perial railroad and communication between 142 00:08:14.639 --> 00:08:17.759 the two point being, to answer your question, if you 143 00:08:17.839 --> 00:08:21.279 go with inertia, there's service any capability. 144 00:08:21.360 --> 00:08:23.439 So if I want to do a Larrabell and view. 145 00:08:23.480 --> 00:08:26.279 Then I can do that like I am seeing. Okay, 146 00:08:26.319 --> 00:08:28.199 to be clear, I'm looking. I opened it up in 147 00:08:28.240 --> 00:08:30.240 a second and I'm showing me some not js. 148 00:08:33.879 --> 00:08:37.080 On Well, it's possible. I haven't used. 149 00:08:37.039 --> 00:08:39.399 It's like it's the way to do these kind of 150 00:08:39.440 --> 00:08:42.799 things typically is there is some no JS little notes 151 00:08:42.879 --> 00:08:45.200 thing running and doing the job. 152 00:08:45.279 --> 00:08:45.720 I don't know. 153 00:08:45.799 --> 00:08:50.759 It somewhere, okay, because in the problem is that with 154 00:08:50.879 --> 00:08:54.919 modern front and frameworks you need the areizomorphic. So you 155 00:08:55.039 --> 00:08:57.639 have the same Java skipt code that runs on the 156 00:08:57.639 --> 00:08:59.600 front and on the back end, and in order to 157 00:08:59.639 --> 00:09:03.960 do some side rendering you have to render it on 158 00:09:03.960 --> 00:09:04.559 on the back. 159 00:09:05.000 --> 00:09:05.559 Okay. 160 00:09:05.799 --> 00:09:08.360 So jovscripture on time on the back end. 161 00:09:08.559 --> 00:09:10.240 And you need a Java sleap run time on the 162 00:09:10.240 --> 00:09:11.840 back okay to. 163 00:09:11.840 --> 00:09:13.960 Do this it's underneath. Yeah, okay, I get you. 164 00:09:14.159 --> 00:09:15.559 So this is why I'm saying. 165 00:09:15.559 --> 00:09:17.320 This is the point of there is always some no 166 00:09:17.440 --> 00:09:21.120 JS everywhere in in In. 167 00:09:21.120 --> 00:09:23.840 And then kind and then PLATFORMA. They kind of flipped 168 00:09:23.840 --> 00:09:26.919 the script there where you kind of put WordPress and 169 00:09:27.000 --> 00:09:28.960 Narvel on top of note or something. 170 00:09:30.360 --> 00:09:33.399 Yes, So last last year we shaped a few experiments. 171 00:09:33.799 --> 00:09:37.639 We were able to run full pitchp inside no JS. 172 00:09:38.840 --> 00:09:42.480 It works really well. It's actually very performant, to be honest. 173 00:09:43.159 --> 00:09:49.360 So there is literally nothing stopped that kind of those 174 00:09:49.440 --> 00:09:55.159 kind of deployments. But it's let's see what the market says, 175 00:09:55.279 --> 00:09:58.720 and if there is, if they are attractive for the market, 176 00:09:58.799 --> 00:10:00.679 we will keep for showing those. 177 00:10:01.360 --> 00:10:03.120 Is it raging or something? 178 00:10:03.519 --> 00:10:04.519 No binaries. 179 00:10:05.120 --> 00:10:09.000 It is the actual PHP right high from the operating system. 180 00:10:09.159 --> 00:10:11.120 It's the same thing. It's it's the same hooks that 181 00:10:11.120 --> 00:10:11.879 the patchy uses. 182 00:10:12.919 --> 00:10:17.240 Hm hmm, Okay, now I understand what you're doing. That's interesting. 183 00:10:19.360 --> 00:10:22.399 So we were talking about benchmarking. Does how does this 184 00:10:22.519 --> 00:10:23.080 tie back in? 185 00:10:23.440 --> 00:10:30.039 Yeah, so we've been doing this is the benchmarks. Okay, 186 00:10:30.360 --> 00:10:32.720 we've done a few. I just wanted to point out 187 00:10:32.919 --> 00:10:38.519 chew one is and let's see where the chat is. 188 00:10:38.559 --> 00:10:40.600 Here is a chat. 189 00:10:41.679 --> 00:10:43.919 Yeah, there's a chat. It's on the right. It's got 190 00:10:43.919 --> 00:10:44.759 a little chat bubble. 191 00:10:44.799 --> 00:10:46.600 Okay, here we go. I'm passing in the chat. So 192 00:10:46.679 --> 00:10:48.879 this is one. I am passing it. 193 00:10:49.200 -->

00:10:57.639 The second one that we have done, and then I 194 00:10:57.679 --> 00:10:58.559 have a few more. 195 00:10:58.399 --> 00:10:59.080 To talk about. 196 00:10:59.240 --> 00:11:02.519 So the first one is we've done last year. It's 197 00:11:02.559 --> 00:11:08.799 a benchmark showing next JS performance. Okay, under very high traffic. 198 00:11:09.919 --> 00:11:13.279 So for Next, yes, so what happened, What is happening 199 00:11:13.600 --> 00:11:20.759 is we have done a lot of tests, okay, and 200 00:11:21.440 --> 00:11:25.519 it's with a lot of tests for example, with one 201 00:11:25.519 --> 00:11:29.360 thousand requests on per second for I mean, for a 202 00:11:29.399 --> 00:11:36.440 couple of minutes on against next JS and those were 203 00:11:36.519 --> 00:11:37.919 hitting essentially saturation. 204 00:11:38.039 --> 00:11:39.320 This was a hellower page. 205 00:11:39.720 --> 00:11:40.039 Okay. 206 00:11:40.519 --> 00:11:45.559 It's very normal for your system to go to if 207 00:11:45.600 --> 00:11:48.840 you're using Next or a React based server side system, 208 00:11:49.120 --> 00:11:53.840 to go between eight twenty, between eight and twenty maybe 209 00:11:53.919 --> 00:11:55.559 fifty requests per second at. 210 00:11:55.440 --> 00:11:56.720 Best, okay. 211 00:11:56.759 --> 00:11:59.279 If you have a very complex system like it's low 212 00:11:59.559 --> 00:12:05.720 okay in the genetic terms of things, okay. So the 213 00:12:06.039 --> 00:12:12.919 critical point okay of this was the success rate, but 214 00:12:13.039 --> 00:12:16.159 more important in the latency okay. The latency for this 215 00:12:16.279 --> 00:12:19.320 kind of system is very bad okay. And this is 216 00:12:19.360 --> 00:12:25.320 affecting a lot of companies out there, okay. And a 217 00:12:25.360 --> 00:12:32.519 good chunk of that problem is due to our scheduling 218 00:12:32.639 --> 00:12:39.679 is done of of multiple parts in kubernets okay, round robin. Unfortunately, 219 00:12:39.799 --> 00:12:43.279 it's it's not the best algorithm most of the time, okay, 220 00:12:44.399 --> 00:12:49.360 But the one that's the one we have. And last, 221 00:12:49.360 --> 00:12:53.200 but not least, there is the problem of event look blocking. 222 00:12:53.799 --> 00:12:55.360 Event look blocking it's critical. 223 00:12:56.039 --> 00:12:58.240 So if the if the event look blocks, then know 224 00:12:59.000 --> 00:13:06.679 the system becomes responsive after this, So it's it's it's problematic. 225 00:13:07.039 --> 00:13:10.200 But then the whole thing about note that the event 226 00:13:10.320 --> 00:13:12.960 loup is kind of never supposed to be blocked. 227 00:13:14.600 --> 00:13:22.039 Well, tell that to all the developers using React, like, 228 00:13:22.320 --> 00:13:26.000 that's the absolutely, that's the world point the idea. Yeah, 229 00:13:26.120 --> 00:13:26.919 that's the idea. 230 00:13:27.600 --> 00:13:35.159 But reacts over side rendering is super

expensive, okay, and 231 00:13:36.320 --> 00:13:37.840 it's super expensive. 232 00:13:37.759 --> 00:13:42.200 So yeah, a lot of string operations and stuff. 233 00:13:42.720 --> 00:13:45.559 No, it's a massive So basically what React does It 234 00:13:45.600 --> 00:13:48.799 creates a full virtual dam of all the elements in 235 00:13:48.840 --> 00:13:51.480 the page and then it stringifies that and you need 236 00:13:51.519 --> 00:13:54.960 to walk through a massive tree for every request. 237 00:13:55.080 --> 00:13:58.120 So this create a massive pressure on the GC as. 238 00:13:58.039 --> 00:14:01.399 Well as the computational cost of a locating of stringifying 239 00:14:01.399 --> 00:14:02.320 all of those elements. 240 00:14:02.639 --> 00:14:06.559 That's really unfortunate given that it then immediately discards that 241 00:14:06.759 --> 00:14:09.480 entire virtual dome exactly. 242 00:14:10.480 --> 00:14:13.279 You know why I've been a pretty like there is. 243 00:14:13.679 --> 00:14:16.200 The only framework that does better is solid I think 244 00:14:16.440 --> 00:14:18.279 is significantly better on the topic. 245 00:14:18.440 --> 00:14:22.000 Okay, maybe because it doesn't have a virtual dome. 246 00:14:23.360 --> 00:14:28.919 This is a good point. This is actually a good point. 247 00:14:29.120 --> 00:14:34.600 Okay, And yeah, so the word point of note is 248 00:14:34.639 --> 00:14:39.639 not blocking the end loop and I can really subscribe 249 00:14:39.679 --> 00:14:43.600 to that sentence. Okay, So to the point, so to 250 00:14:43.639 --> 00:14:47.519 the point that if the event look blocks for a 251 00:14:47.559 --> 00:14:51.600 long time, it's better to kill that node process and 252 00:14:51.639 --> 00:14:57.000 start a fresh one up because keeping that running is useless. 253 00:14:57.799 --> 00:15:02.679 Yeah, if you're going to be doing really significant GC, 254 00:15:04.000 --> 00:15:07.600 then why I mean, just you know, I'm reminded of 255 00:15:07.679 --> 00:15:13.240 what the Apache I think the approach was about, like 256 00:15:13.480 --> 00:15:19.039 managing like the heap in in like sort of chunks 257 00:15:19.039 --> 00:15:21.759 and then discarding the entire chunk rather than trying to 258 00:15:21.759 --> 00:15:23.200 do fun grain GC. 259 00:15:24.519 --> 00:15:28.960 Well, it's to some extent is relatively similar what we 260 00:15:28.960 --> 00:15:31.799 are doing. So with what or not just application server 261 00:15:32.919 --> 00:15:35.600 we have based on this on this thing, we can 262 00:15:35.679 --> 00:15:37.919 monitor the state of the event loop from the outside 263 00:15:38.320 --> 00:15:42.080 and if you detect it, it is totally blocked. Nothing comes true, 264 00:15:42.120 --> 00:15:44.440 nothing is going to come through. We just shut the 265 00:15:44.720 --> 00:15:48.799 shut the thing down. So and because we can, we 266 00:15:48.879 --> 00:15:52.039 shut the thing down and we started from and restart, 267 00:15:52.240 --> 00:15:57.120 we can essentially recover a really broken pad back to 268 00:15:57.240 --> 00:15:59.240 life and let some traffic pass through. 269 00:15:59.480 --> 00:16:04.159 So the basic idea is saying, instead of trying to 270 00:16:04.240 --> 00:16:08.879 do this really really complex GCS at a high frequency, 271 00:16:09.000 --> 00:16:12.919 because of all the incoming requests, every once in a while, 272 00:16:13.039 --> 00:16:15.679 just kill the pod and start fresh. 273 00:16:15.720 --> 00:16:18.559 Well, it's when the event will be is blocked. So 274 00:16:18.759 --> 00:16:22.120 it's literally like when there are too many like. The 275 00:16:22.159 --> 00:16:24.200 problem is also that there are all those requests are 276 00:16:24.200 --> 00:16:28.399 piling up to be processed and after five seconds it's 277 00:16:28.440 --> 00:16:28.919 no point. 278 00:16:29.600 --> 00:16:32.440 There's no point in processing them. You' just abort them. 279 00:16:32.759 --> 00:16:36.240 And it's because of the GC, because of the virtual dome. 280 00:16:36.279 --> 00:16:37.399 It's all of both of them. 281 00:16:37.840 --> 00:16:40.879 It's it's a very it's a very superre intensive operation, 282 00:16:41.080 --> 00:16:42.039 especially with a lot. 283 00:16:41.879 --> 00:16:47.440 Of interesting You would think that they would have figured 284 00:16:47.440 --> 00:16:50.840 it out a way to offload it to workers. 285 00:16:50.440 --> 00:16:52.639 Or something, which is what we did. 286 00:16:53.440 --> 00:16:55.639 Is that a different thing or the no. 287 00:16:55.840 --> 00:17:00.480 So basically with what we can have multiple workers listening 288 00:17:00.519 --> 00:17:05.200 to the same socket so that you can do the 289 00:17:05.240 --> 00:17:08.200 load balancing easier. It's very similar to say, oh, it 290 00:17:08.240 --> 00:17:11.799 is similar to nod cluster or PM two and not cluster, 291 00:17:12.119 --> 00:17:16.359 but instead of doing the load balancing inside not JS, 292 00:17:16.960 --> 00:17:21.240 which is what nod cluster and pmhould do with in 293 00:17:21.400 --> 00:17:24.920 that we use a new node feature that let us 294 00:17:25.000 --> 00:17:29.039 do load balancing inside the kernel using scores, a flag 295 00:17:29.119 --> 00:17:34.000 calls so on the score reuse port, which is significantly faster. 296 00:17:34.920 --> 00:17:38.319 Can you elaborate a little bit about what actually is? 297 00:17:38.759 --> 00:17:40.880 Because it's an application showy for not JS, you can 298 00:17:40.960 --> 00:17:42.799 just give it your app. It will run it inside 299 00:17:42.799 --> 00:17:45.240 a worker thread and the main thread will just minor 300 00:17:45.359 --> 00:17:48.359 it and you can scale it up and down multiple threads, 301 00:17:48.400 --> 00:17:51.960 multiple course, and if you want, you can even have 302 00:17:52.039 --> 00:17:55.519 aterogenerus system. You can have multiple different

applications running there 303 00:17:55.720 --> 00:17:58.839 and they can communicate via message passing or just h GDP 304 00:17:59.400 --> 00:18:00.000 over memory. 305 00:18:00.079 --> 00:18:04.039 See so what does it compete with? What are they like? 306 00:18:05.519 --> 00:18:08.039 When would I use what? And would I put like 307 00:18:08.279 --> 00:18:10.160 next JS? On top of what? 308 00:18:10.720 --> 00:18:13.440 If you're if you're deploying not JES in Kubernetus, you're 309 00:18:13.480 --> 00:18:16.759 probably not doing the not doing the right thing if 310 00:18:16.799 --> 00:18:18.839 you're not doing it today using what? 311 00:18:19.640 --> 00:18:22.279 Yes, what does it stand for? By the way? 312 00:18:23.279 --> 00:18:27.400 What come on? What can what be? Can that be? 313 00:18:28.319 --> 00:18:28.599 Come on? 314 00:18:29.000 --> 00:18:30.480 How do you exactly spell it? 315 00:18:31.440 --> 00:18:36.160 W A T T. It's an honor to over James vat. 316 00:18:36.640 --> 00:18:44.640 Ah Okay, yeah, the power. Yeah, so we need a name. 317 00:18:44.720 --> 00:18:46.680 So on MPM is VATPMH. 318 00:18:46.680 --> 00:18:49.359 I now I sit in your blog post. Yes, so 319 00:18:49.440 --> 00:18:53.480 whenever I use node in Kubernetes, you recommend using it 320 00:18:53.519 --> 00:18:55.160 on top of what I. 321 00:18:55.200 --> 00:18:59.079 Would say, Yes, it gives you more reliability in this 322 00:18:59.240 --> 00:19:03.599 example with jes that I've shared in the chat, essentially 323 00:19:03.640 --> 00:19:08.000 we can move down P ninety five from a second 324 00:19:08.720 --> 00:19:11.920 to two one hundred and thirty five milliseconds. 325 00:19:12.440 --> 00:19:15.640 Oh wow. And just for those who don't follow along 326 00:19:15.680 --> 00:19:19.640 with benchmark stuff of P ninety five, is ninety five 327 00:19:19.720 --> 00:19:23.000 percent of your requests take this long or less? 328 00:19:23.400 --> 00:19:28.000 So in other words, yeah, beneficial for the five percent 329 00:19:28.119 --> 00:19:29.359 slowest users. 330 00:19:29.400 --> 00:19:31.000 Is beneficial for everybody. 331 00:19:30.799 --> 00:19:34.960 If fits for everybody. Yeah, essentially, Yeah, so all the 332 00:19:35.160 --> 00:19:38.960 all the requests now, so ninety five percent of the 333 00:19:39.079 --> 00:19:41.759 requests instead of taking up to a second, now take 334 00:19:42.039 --> 00:19:44.559 up to two hundred and thirty five milliseconds. Now you 335 00:19:44.599 --> 00:19:46.960 assume that the other five percent it also speeds up 336 00:19:47.000 --> 00:19:51.039 for But yeah, that's kind of the measure, and it's 337 00:19:51.079 --> 00:19:53.319 it's one of the more common ways of measuring how 338 00:19:53.319 --> 00:19:57.680 fast your application is because the load and the requests 339 00:19:57.720 --> 00:19:59.640 and what it has to do and what else is 340 00:19:59.759 --> 00:20:01.880 running on the machine can vary from time to time,

341 00:20:01.920 --> 00:20:03.519 and so this kind of averages it out. 342 00:20:03.880 --> 00:20:07.680 Also, think about it from the perspective of a business 343 00:20:08.240 --> 00:20:13.000 loto you like, you want to think about the customers 344 00:20:13.000 --> 00:20:16.359 that have it the worst, and you can't really afford 345 00:20:16.400 --> 00:20:19.119 to lose five or ten percent of your business because 346 00:20:19.200 --> 00:20:26.680 it's just too slow. Exactly. But I'm thinking about what 347 00:20:26.720 --> 00:20:30.759 you said. It does mean though, that you want to 348 00:20:30.839 --> 00:20:36.640 allocate more than one CPU per yes, no instance in 349 00:20:36.680 --> 00:20:38.079 your kubernets in how. 350 00:20:37.960 --> 00:20:42.839 You absolutely this actually helps. Okay, let me explain why 351 00:20:42.880 --> 00:20:43.319 it helps. 352 00:20:43.480 --> 00:20:48.079 Okay, The cost of spinning up a new pod in 353 00:20:48.200 --> 00:20:54.039 Kubernatis is more or less fixed independent of the amount 354 00:20:54.359 --> 00:21:01.279 the amount of resources that are attached, okay, or anyway, 355 00:21:01.279 --> 00:21:02.720 it's not very much different. 356 00:21:03.240 --> 00:21:03.519 Okay. 357 00:21:04.359 --> 00:21:08.680 If I am scheduling largest things, each unit of scale 358 00:21:08.880 --> 00:21:14.079 gives me more power to handle my spikes better. So 359 00:21:14.559 --> 00:21:17.599 it's the edges of scaling are a little bit steeper. 360 00:21:18.279 --> 00:21:23.240 But this actually helps smooth the curve way quicker because 361 00:21:23.359 --> 00:21:24.960 these new things. 362 00:21:26.519 --> 00:21:27.119 Make it. 363 00:21:28.720 --> 00:21:32.640 Can handle more load per unit. Okay, So this is 364 00:21:32.640 --> 00:21:33.119 one thing. 365 00:21:33.279 --> 00:21:33.640 Okay. 366 00:21:33.680 --> 00:21:37.960 The second thing is Kubernatis takes minutes to scale, so 367 00:21:38.079 --> 00:21:41.400 when the load typically reacts in one two minutes is 368 00:21:41.480 --> 00:21:44.960 very slow, tiny to to to schedule things up. And 369 00:21:46.039 --> 00:21:49.480 by having this system you can actually absorb the shock 370 00:21:49.680 --> 00:21:52.200 completely inside your pods. 371 00:21:52.200 --> 00:21:55.279 Interesting, very interesting. I need to think about this some more. 372 00:21:55.359 --> 00:21:59.279 I will definitely make sure that our develops people see this. 373 00:22:02.759 --> 00:22:06.200 I mean literally everybody is using kubernets these days, whether 374 00:22:06.200 --> 00:22:06.640 they need. 375 00:22:06.559 --> 00:22:08.720 To exactly exactly. 376 00:22:09.359 --> 00:22:12.480 We also have built, but it's not mentioned in this benchmark, 377 00:22:12.720 --> 00:22:17.279 built another product called Intelligent Command Center that is open 378 00:22:17.319 --> 00:22:23.119 source as well, that reduced the decision time of kubernets 379 00:22:23.720 --> 00:22:32.279

from minutes two seconds. So the pod starting is down 380 00:22:32.480 --> 00:22:35.200 is making less than five seconds the decision to start 381 00:22:35.200 --> 00:22:35.680 a new part. 382 00:22:36.400 --> 00:22:39.279 Doesn't that make the system kind of more noisy. 383 00:22:40.200 --> 00:22:43.759 Yes to some part, not to the other, because it's 384 00:22:43.799 --> 00:22:47.799 actually it's actually very fast in scaling up. Okay, but 385 00:22:47.920 --> 00:22:51.920 it can scale up on very specific signals from the app, 386 00:22:52.640 --> 00:22:58.200 so it and so it can be a little bit 387 00:22:58.200 --> 00:23:00.839 more noisy, but it actually allows you to reduce your 388 00:23:00.880 --> 00:23:03.359 baseline of instance significantly. 389 00:23:03.279 --> 00:23:06.000 So you can scale up when you need to. 390 00:23:06.359 --> 00:23:08.319 You can scale up way quicker when you need to, 391 00:23:08.359 --> 00:23:11.559 so you can reduce your baseline and during off peak 392 00:23:11.599 --> 00:23:15.960 hours you can actually save a lot of Yeah. 393 00:23:15.799 --> 00:23:18.480 I guess that's always the thing that I'm concerned with 394 00:23:18.519 --> 00:23:21.480 when you have this kind of automation, is, you know, 395 00:23:21.559 --> 00:23:24.400 does a scale up too much? Does it? You know, 396 00:23:24.400 --> 00:23:26.319 does it handle things when you know when I need 397 00:23:26.319 --> 00:23:28.559 it to scale down? Because yeah, it costs a lot 398 00:23:28.599 --> 00:23:31.440 more when you scale way up. But if that's what 399 00:23:31.480 --> 00:23:35.680 you're getting in traffic, and usually traffic translates in some 400 00:23:35.720 --> 00:23:40.559 ways to money, not always, but yeah, so that anyway 401 00:23:41.240 --> 00:23:41.839 pretty cool. 402 00:23:42.000 --> 00:23:45.079 So this was part of our results. 403 00:23:45.119 --> 00:23:51.279 We published this study, okay, and hopefully we can get 404 00:23:51.359 --> 00:23:54.279 some results from our rock click customer down out soon. 405 00:23:54.799 --> 00:23:58.640 But this is our own and all the benchmarkt and 406 00:23:58.759 --> 00:24:01.279 ernests done for this is open source, so you can 407 00:24:01.359 --> 00:24:02.759 check it out and run it yourself. 408 00:24:02.799 --> 00:24:03.440 If you're this. 409 00:24:03.400 --> 00:24:06.559 Should be making a lot more noise than it is. 410 00:24:06.640 --> 00:24:10.119 I think I was kind of aware of this, but 411 00:24:10.200 --> 00:24:13.359 I was not fully aware of this. And like I said, 412 00:24:13.400 --> 00:24:17.759 literally everybody is doing non kumberinators these days. So if 413 00:24:17.799 --> 00:24:22.039 the impact is so dramatic and so significant and so 414 00:24:22.519 --> 00:24:29.599 readily achievable, yes, you know, there's literally no reason not 415 00:24:29.720 --> 00:24:32.079 to try it out. This is. 416 00:24:33.920 --> 00:24:36.359 Partially the reason why when you asked to

come to 417 00:24:36.440 --> 00:24:38.480 join the show. I said, I'm coming join the show. 418 00:24:39.839 --> 00:24:44.680 This is the research that we published. So we published 419 00:24:44.759 --> 00:24:49.079 late last year, I think, and it's pretty solid. 420 00:24:49.400 --> 00:24:57.359 I would say. Then we have done another article. 421 00:24:57.599 --> 00:25:03.599 Okay, I'm more spicy one if you if a more 422 00:25:03.640 --> 00:25:09.160 spicy one, I would say, okay. And we benchmarked on 423 00:25:09.200 --> 00:25:11.839 the same benchmark on the first one, slightly different traffic 424 00:25:11.880 --> 00:25:17.720 load whatever, Okay. We benchmarked the note the tree no 425 00:25:17.839 --> 00:25:22.880 dress run times in Kubernetes so bunn no JS as 426 00:25:22.880 --> 00:25:25.519 well as our own way of scheduling things with what 427 00:25:26.200 --> 00:25:33.640 so we got some really unexpected results, not the results 428 00:25:33.640 --> 00:25:41.640 that I would expect, and it showed that bun So. 429 00:25:41.720 --> 00:25:45.000 Just before you you tell what you found, just to clarify, 430 00:25:45.200 --> 00:25:48.519 what is only for node, it's not for the other 431 00:25:48.960 --> 00:25:51.079 For the other platforms, you can't use it to DNO 432 00:25:51.240 --> 00:25:51.960 with bun. 433 00:25:52.039 --> 00:25:53.039 No you can't. 434 00:25:53.519 --> 00:25:57.519 Well like you can't because those other platforms are not 435 00:25:57.680 --> 00:26:01.200 fully not GS compatible. And the approach that we have 436 00:26:01.279 --> 00:26:05.440 taken with what is instead of creating a custom built 437 00:26:05.599 --> 00:26:08.519 not just run time with our patches, we have upstream 438 00:26:08.519 --> 00:26:11.839 all our patches up to note core of the things 439 00:26:11.839 --> 00:26:17.319 that we needed, and that is only is written mostly 440 00:26:17.319 --> 00:26:22.119 in JavaScript mostly every now and then we had to 441 00:26:22.200 --> 00:26:25.920 have a little bit of C plus plus to implement 442 00:26:25.960 --> 00:26:28.480 some feature for all not Jazz versions. 443 00:26:28.160 --> 00:26:30.079 That are not being released yet. 444 00:26:30.160 --> 00:26:34.200 So if we need a feature, we need to extend 445 00:26:34.200 --> 00:26:34.880 support for it. 446 00:26:35.039 --> 00:26:39.759 But it's right now, I think it's all JavaScript. 447 00:26:40.079 --> 00:26:42.119 I have to say before you go too much further 448 00:26:42.200 --> 00:26:45.759 that I like this idea of hey, we needed this 449 00:26:45.880 --> 00:26:48.799 in node in order to make what work, and so 450 00:26:48.839 --> 00:26:50.920 you contribute it back and so it's stuff that other 451 00:26:50.920 --> 00:26:51.880 people can use. 452 00:26:52.119 --> 00:26:56.039 Exactly one like this is the whole point. Like part 453 00:26:56.079 --> 00:26:59.680 of the reason why I funded Platformatic was can I 454 00:26:59.759 --> 00:27:06.240 try to

build a business that can uh make the 455 00:27:06.440 --> 00:27:11.960 open source economically viable and and allow us to grow 456 00:27:12.000 --> 00:27:15.480 with it? So let's try. This is the this might 457 00:27:15.519 --> 00:27:19.200 try Okay, I might succeed, might not succeed. I hope 458 00:27:19.200 --> 00:27:24.400 I succeed, but I needed to try this. So my journey. 459 00:27:25.000 --> 00:27:26.759 I'm going to go into a tangent here real quick 460 00:27:26.759 --> 00:27:28.880 because I'm kind of curious. So when you say make 461 00:27:29.079 --> 00:27:33.000 the open source uh economically viable, what do you mean? 462 00:27:33.079 --> 00:27:33.160 Like? 463 00:27:33.240 --> 00:27:35.160 What what the fundamental point is? 464 00:27:35.359 --> 00:27:45.160 The fundamental point is there is a substantial lack of 465 00:27:45.599 --> 00:27:53.359 investment in core technologies from companies, and we need more 466 00:27:53.440 --> 00:27:59.200 companies that are that invest significant amount of money into 467 00:27:59.240 --> 00:27:59.880 the commons. 468 00:28:00.960 --> 00:28:03.160 In order to do so, we need them. 469 00:28:03.000 --> 00:28:03.319 To be. 470 00:28:04.799 --> 00:28:08.319 Economic, to be good businesses so they can contribute back 471 00:28:08.519 --> 00:28:14.920 and up. Okay, And you cannot rely on the sponsor 472 00:28:15.000 --> 00:28:20.440 model because it's too little, too late. So you need 473 00:28:21.359 --> 00:28:23.839 if you need to pay to do payroll, you need 474 00:28:23.920 --> 00:28:29.960 to have significant sales involved. So the goal is to 475 00:28:30.000 --> 00:28:33.680 build a significantly big enough company that can afford multiple 476 00:28:33.680 --> 00:28:36.559 people and working full time on the core technologies, and 477 00:28:36.759 --> 00:28:39.880 not because it's it's a good thing to do, but 478 00:28:40.119 --> 00:28:44.440 because it needs them to be there as a core 479 00:28:44.559 --> 00:28:47.640 function of the business so that it's not a donation, 480 00:28:47.960 --> 00:28:49.480 it's a function. 481 00:28:50.559 --> 00:28:51.480 Okay. 482 00:28:51.599 --> 00:28:54.640 And because it's a function and it's a core part 483 00:28:54.640 --> 00:28:56.359 of it, it will not stop. 484 00:28:57.400 --> 00:28:59.440 And what is that company selling them? 485 00:29:00.359 --> 00:29:02.000 That was what I wash. 486 00:29:02.079 --> 00:29:07.079 So our current business model okay, after several pivot iterations 487 00:29:07.079 --> 00:29:13.519 and stuff, we sell enterprise version of all our technologies 488 00:29:13.759 --> 00:29:21.039 plus support and on boarding. So if somebody needs anything 489 00:29:21.079 --> 00:29:24.039 related to no JS, we can provide support, We can 490 00:29:24.200 --> 00:29:30.599 provide and we can provide an enterprise license for our 491 00:29:30.640 --> 00:29:32.559 core technologies and include. 492 00:29:32.279 --> 00:29:33.799 What does an enterprise license. 493 00:29:33.880 --> 00:29:39.559 Mean this is an answer for Michael question for micro founder. Okay, 494 00:29:41.400 --> 00:29:44.839 so there is for the Intelligent Command Center. 495 00:29:45.480 --> 00:29:46.519 There is a. 496 00:29:48.480 --> 00:29:56.319 Premium version with added features for example like more audit controls, loggins, 497 00:29:56.519 --> 00:30:00.440 these kind of things so far, but it will grow more. 498 00:30:00.880 --> 00:30:08.119 It's more related to SLAS and response to incidents and 499 00:30:08.599 --> 00:30:10.200 being available that. 500 00:30:11.799 --> 00:30:13.440 Like other things. 501 00:30:13.839 --> 00:30:16.319 So it's more on the support side than the future side. 502 00:30:17.039 --> 00:30:20.079 It's the future side is more on all these technology 503 00:30:20.119 --> 00:30:26.839 are moving dynamically, so they our customers can steer, can 504 00:30:26.839 --> 00:30:29.119 steer the wheel and I'll pass the side. 505 00:30:29.160 --> 00:30:30.000 Where we go next? 506 00:30:30.079 --> 00:30:32.680 And what what they need? What features are missing? And 507 00:30:33.119 --> 00:30:34.480 you know their partners. 508 00:30:34.359 --> 00:30:36.759 I think that esslas are the way to go, to 509 00:30:36.799 --> 00:30:38.079 be honest, To. 510 00:30:38.000 --> 00:30:39.920 Be honest, I would say they sell very. 511 00:30:39.759 --> 00:30:44.440 Well because like literally, if I'm running your back in 512 00:30:44.519 --> 00:30:49.920 my back end on node, I need ESLAS and and 513 00:30:50.720 --> 00:30:54.799 if you know, if I if I can out kind 514 00:30:54.839 --> 00:31:01.400 of guaranteed by licensing your services, then it's no brainer. 515 00:31:01.799 --> 00:31:07.160 That's what it is. So anyway, that's exactly the point. 516 00:31:07.559 --> 00:31:10.680 Cool. So now let's go back to where we were 517 00:31:10.720 --> 00:31:12.079 before we went on the tangent. 518 00:31:12.160 --> 00:31:14.559 If we can remember, yeah, okay, we can go on 519 00:31:14.599 --> 00:31:20.319 the tangent all the time. So anyway we are, I 520 00:31:20.359 --> 00:31:24.200 was chatting, I've over the break, I've done this extensive 521 00:31:24.200 --> 00:31:30.240 benchmark on band Dino and note okay, and I got 522 00:31:30.519 --> 00:31:34.319 very surprising results. The benchmarks included running the same next 523 00:31:34.359 --> 00:31:35.400 Jaess application that. 524 00:31:35.400 --> 00:31:39.400 We talked about before. Okay, reactions are rendering? Have your memory? 525 00:31:39.480 --> 00:31:40.559 Have you on CPU? 526 00:31:41.039 --> 00:31:41.920 Really like. 527 00:31:44.039 --> 00:31:47.880 Possibly the worst load but also the most common load 528 00:31:48.400 --> 00:31:52.599 that you can have running on OJS these days, And 529 00:31:52.680 --> 00:31:55.400 I got very surprising result that I didn't expect them 530 00:31:55.440 --> 00:31:59.519 to be so specifically. 531

00:31:59.480 --> 00:32:06.039 I got but p ninety nine of. 532 00:32:08.519 --> 00:32:12.319 Seventy four milliseconds for ban one three dot five they 533 00:32:12.319 --> 00:32:13.519 did new versions of. 534 00:32:13.839 --> 00:32:15.279 So approximately one second. 535 00:32:16.279 --> 00:32:20.839 Yes, well not JS was one hundred and seventy four. 536 00:32:21.799 --> 00:32:24.960 Do you know at one hundred and not JS with 537 00:32:25.039 --> 00:32:26.640 what at one hundred and fifteen? 538 00:32:27.480 --> 00:32:33.599 So Bun was noticeably slower than everybody else exactly. To 539 00:32:33.680 --> 00:32:37.240 be honest, I know that Bun has a reputation for 540 00:32:37.319 --> 00:32:40.799 being fast, but the people need but people need to 541 00:32:40.880 --> 00:32:44.920 understand that fast can mean different things in different situations. 542 00:32:45.480 --> 00:32:46.359 Exactly that. 543 00:32:47.000 --> 00:32:49.680 Well, I hear people tell other things about Bun too, 544 00:32:49.680 --> 00:32:54.799 but yeah, you run a different application the different engines 545 00:32:54.920 --> 00:32:59.720 manage memory differently and concurrency differently than other things differently, 546 00:32:59.720 --> 00:33:01.039 and so yeah, it's. 547 00:33:00.920 --> 00:33:04.079 Beyond a different application. I think it's it's like a 548 00:33:04.079 --> 00:33:08.799 different scenario, yes, because you've got you've got startup times, 549 00:33:09.440 --> 00:33:13.799 and you've got and you've got run times, and there's 550 00:33:13.799 --> 00:33:17.400 a question of when optimizations kick in. If they kick in, 551 00:33:18.440 --> 00:33:20.960 what do you do in the jet stage stuff like that. 552 00:33:21.240 --> 00:33:25.559 So what could be really fast for a long running 553 00:33:26.759 --> 00:33:30.640 a no type service could be really slow when you're 554 00:33:30.680 --> 00:33:35.440 building a utility that just runs, does something and then terminates, 555 00:33:35.759 --> 00:33:40.960 and an expectation that a certain platform will do everything 556 00:33:41.039 --> 00:33:46.839 better is well, it's happened before, but it's not always realistic. 557 00:33:47.559 --> 00:33:50.119 It's so a few notes. 558 00:33:50.279 --> 00:33:53.799 Okay, after chatting with a few people, they gave me 559 00:33:53.880 --> 00:33:57.200 this informal feedback. You can take what we do with 560 00:33:57.279 --> 00:34:03.200 a gain of salt jsc which is the JavaScript run 561 00:34:03.279 --> 00:34:08.199 time behind ban and Safari is optimized for cold starts 562 00:34:08.800 --> 00:34:11.880 to start very immediately. Imagine when you open Safari on 563 00:34:11.920 --> 00:34:14.159 your iPhone, it opens up super quick. 564 00:34:14.480 --> 00:34:22.599 Okay, while V eight is more optimized for running or 565 00:34:22.639 --> 00:34:24.719 wrong running application. Thanks Jmail. 566 00:34:25.519 --> 00:34:28.519 Yeah, that's the thing, you know, people

talk about like 567 00:34:28.639 --> 00:34:33.159 the various You know, there's this been trend of rewriting 568 00:34:33.239 --> 00:34:37.719 the JavaScript utility ecosystem in other programming languages, be it 569 00:34:37.840 --> 00:34:42.039 go or be at rust or be it whatever, instead 570 00:34:42.079 --> 00:34:47.280 of doing it in JavaScript or typescript. And you kind 571 00:34:47.280 --> 00:34:49.880 of hit your head on the nail there, matel Because, 572 00:34:49.920 --> 00:34:51.360 for example, if we think about. 573 00:34:51.159 --> 00:34:53.360 And that's a nail on the head, not head on 574 00:34:53.360 --> 00:34:55.079 the nail unless you really want to hit your head 575 00:34:55.119 --> 00:34:55.599 on the nail. 576 00:34:55.800 --> 00:34:57.960 Or is that what I said? Yeah, okay, just. 577 00:34:58.159 --> 00:34:59.320 To point ound was painful. 578 00:34:59.599 --> 00:35:03.280 Well know Hebrew we right, we right right to left. 579 00:35:03.360 --> 00:35:07.039 So well, but I mean, gives gives new gives new 580 00:35:07.079 --> 00:35:08.639 meaning to the term hammerhead, doesn't it. 581 00:35:09.000 --> 00:35:13.119 Yeah? Anyway, what I was saying is that the JavaScript 582 00:35:13.199 --> 00:35:17.760 optimizer in EA, I always forget the name whether is 583 00:35:17.840 --> 00:35:18.920 a turbo fan or what. 584 00:35:18.880 --> 00:35:22.639 A turbo fan. We also have mag Lab now. 585 00:35:22.400 --> 00:35:27.360 But yeah, it's basically based on identifying hot code and 586 00:35:27.400 --> 00:35:31.960 then optimizing that while the code is actually executing. But 587 00:35:32.079 --> 00:35:35.159 in order to identify hot code, the application needs to 588 00:35:35.199 --> 00:35:37.719 be running for a while. And if you're running a 589 00:35:37.760 --> 00:35:40.800 tool that just starts up, does its job and then finishes, 590 00:35:41.679 --> 00:35:44.679 you'll never get to that optimization stage. 591 00:35:45.880 --> 00:35:52.280 Exactly, which is long enough you get into that optimization state. 592 00:35:52.400 --> 00:35:54.400 Which is kind of my concern by the way, with 593 00:35:54.920 --> 00:35:57.000 some Lambda function usage. 594 00:35:58.559 --> 00:36:03.119 Well, I let me say this, if you are moving 595 00:36:03.400 --> 00:36:12.239 serious traffic, you are probably better served using ECS or 596 00:36:12.440 --> 00:36:17.519 cobnetis or whatever long running process. But to be honest, 597 00:36:17.599 --> 00:36:20.760 Lambda actually okay, let's let's go. Let's open the lambda 598 00:36:20.800 --> 00:36:24.519 point because it's actually a validation of that to some extent. 599 00:36:25.079 --> 00:36:28.079 So if you if you have followed what has been 600 00:36:28.159 --> 00:36:31.000 launched by a w as in the latest the latest 601 00:36:31.039 --> 00:36:36.559 rainvent in December, is a new Lambda run time for 602 00:36:37.159 --> 00:36:41.239 I troueput scenario I traffic scenario that is not actually 603 00:36:42.280 --> 00:36:46.320 one request per process, okay, but it allows one process 604 00:36:46.480 --> 00:36:49.440 look at this to have as many threads as it 605 00:36:49.519 --> 00:36:52.960 wants to end or requests in a lot of things. 606 00:36:53.119 --> 00:36:58.679 Is actually similar to how that works internally in how 607 00:36:58.840 --> 00:37:03.480 the things are scheduled. So this kind of system allows 608 00:37:04.079 --> 00:37:04.719 Lambda to. 609 00:37:06.559 --> 00:37:09.400 You know, use all the potential of no js essentially 610 00:37:10.119 --> 00:37:17.119 and actually leverage these long running optimizations that we talked about. 611 00:37:17.800 --> 00:37:18.320 Interesting. 612 00:37:19.519 --> 00:37:23.159 So it's actually very interesting model, and a lot of 613 00:37:23.199 --> 00:37:26.079 people are not you know, tapped into this yet, like 614 00:37:26.079 --> 00:37:29.960 they think, oh, no, jes single treded or mostly single traded, 615 00:37:30.639 --> 00:37:33.159 while in reality it's not, and you can do a 616 00:37:33.199 --> 00:37:37.239 lot of interesting things with it, and they just you know, 617 00:37:37.480 --> 00:37:38.320 ignore the point. 618 00:37:38.480 --> 00:37:42.159 Well to be to be honest, no, JavaScript is a 619 00:37:42.159 --> 00:37:47.360 single threaded language, but Node and the JavaScript Virtual Machine 620 00:37:47.360 --> 00:37:49.280 has never really been single threaded. 621 00:37:51.280 --> 00:37:54.840 So it's exactly so for. 622 00:37:54.840 --> 00:37:57.800 Those of you who don't know. For example, the VGC 623 00:37:58.079 --> 00:38:02.440 actually runs off of the main thread to significant extent. 624 00:38:03.320 --> 00:38:06.079 Yeah, well it has been, not at the beginning. 625 00:38:06.280 --> 00:38:08.920 Not at the beginning I'm talking about currently. 626 00:38:09.039 --> 00:38:11.360 Yeah, yeah, it's it's it's off the of it. 627 00:38:12.039 --> 00:38:14.119 So anyway, I just wanted to say we got these 628 00:38:14.360 --> 00:38:18.039 incredible results, like not expected, to be honest, I didn't, 629 00:38:18.280 --> 00:38:20.440 like I had to check them five times and have 630 00:38:20.559 --> 00:38:22.800 to have some friends check them with me. 631 00:38:23.440 --> 00:38:26.400 So why was it expected? You you didn't expect such 632 00:38:26.440 --> 00:38:30.880 a band to be that slow in that particular test, yes, 633 00:38:31.119 --> 00:38:33.880 And did you expect Dino to be that fast by the. 634 00:38:33.880 --> 00:38:38.679 Way, yes, yes, significantly. I was expecting Dino to have 635 00:38:38.719 --> 00:38:39.840 those kind of numbers. 636 00:38:40.320 --> 00:38:44.320 I think that's a scenario that they've been optimizing for, right, No, no. 637 00:38:44.079 --> 00:38:50.519 No, it's the way so they their dino event loop 638 00:38:51.239 --> 00:38:55.199 is based on Tokyo, while no event loop is based 639 00:38:55.239 --> 00:38:55.800 on libuv. 640

00:38:56.280 --> 00:38:58.440 Okay, and Tokyo is. 641 00:39:00.159 --> 00:39:04.400 Be a decade later than libv type of thing, and 642 00:39:04.920 --> 00:39:08.719 by things have improved, you now you do those kinds 643 00:39:08.719 --> 00:39:13.320 of things, okay, And it's very hard to chase something 644 00:39:13.400 --> 00:39:18.000 like libuv understood, So there is that is the problem, 645 00:39:18.199 --> 00:39:20.199 Like it's it's not just LIBUV by the way, it's 646 00:39:20.239 --> 00:39:24.239 also the node machinery in entirety. So how we get 647 00:39:24.280 --> 00:39:27.719 the data out of libuv, how we schedule new connections 648 00:39:27.719 --> 00:39:29.079 and we do all Like there's a lot of work 649 00:39:29.119 --> 00:39:31.679 over there and a lot of over it, but some 650 00:39:31.800 --> 00:39:35.800 of that over it is essentially legacy, and it's very 651 00:39:35.880 --> 00:39:38.320 hard for us to like it's possible, it would be 652 00:39:38.360 --> 00:39:42.000 possible to optimize, but it's a hell of a lot harder. 653 00:39:42.280 --> 00:39:45.679 So could you remind me again what the numbers are 654 00:39:45.760 --> 00:39:47.599 for node versus note with what. 655 00:39:48.599 --> 00:39:49.639 So on p. 656 00:39:49.760 --> 00:39:53.199 Nineteen nine we add one hundred and seventy four millisecond 657 00:39:53.239 --> 00:39:55.320 I may we measured one hundred and seventy four million 658 00:39:55.360 --> 00:39:59.039 second with node and with what one hundred and fifteen. 659 00:40:00.000 --> 00:40:02.719 One hundred and fifteen that's one one five. 660 00:40:03.199 --> 00:40:06.079 Yep, one seven four versus one one five. 661 00:40:07.239 --> 00:40:11.760 So we're talking about close about thirty percent reduction, forty 662 00:40:11.760 --> 00:40:13.239 percent reductions. 663 00:40:12.719 --> 00:40:13.280 Something like that. 664 00:40:13.360 --> 00:40:15.159 Yeah, that's significant. 665 00:40:15.920 --> 00:40:19.880 I know last Dino VAT and Dino are very close. 666 00:40:19.960 --> 00:40:21.559 And that's the SSR scenario. 667 00:40:22.039 --> 00:40:24.000 Yes, by the way, So I guess. 668 00:40:23.800 --> 00:40:26.159 What I'm wondering is is so for the people watching 669 00:40:26.239 --> 00:40:30.480 at home right this, you're throwing out numbers and scenarios, 670 00:40:30.519 --> 00:40:33.639 But what does this look like for me? Right? So 671 00:40:33.719 --> 00:40:36.079 I have a next JS setup, or i have a 672 00:40:36.840 --> 00:40:39.079 note application that I've written, and I've got React on 673 00:40:39.079 --> 00:40:41.719 the front end and I've got server side rendering setup 674 00:40:41.920 --> 00:40:46.079 or whatever, like, how does this actually translate? And what 675 00:40:46.119 --> 00:40:50.239 does this look like when I'm troubleshooting or you know, 676 00:40:50.480 --> 00:40:52.480 measuring my own benchmarks or things like that. 677 00:40:53.079 --> 00:40:58.360 So the key difference is the point of so when 678 00:40:58.360 --> 00:41:02.039 you're doing in your is when you do saturation or 679 00:41:02.079 --> 00:41:07.280 stress testing. Stress testing, so you're putting it to a 680 00:41:07.400 --> 00:41:11.119 very high load scenario and you want to know when 681 00:41:11.119 --> 00:41:12.440 your up breaks. 682 00:41:13.360 --> 00:41:15.440 And essentially. 683 00:41:16.400 --> 00:41:20.960 These numbers tells create more or less ranking of where 684 00:41:21.000 --> 00:41:25.519 your applications stay responsive and is usable by the end user. 685 00:41:25.840 --> 00:41:28.639 Okay, in those scenarios, I'll. 686 00:41:28.400 --> 00:41:32.599 Give another way to look at it. You're doing SSR 687 00:41:33.679 --> 00:41:39.400 to improve application startup times. Yes, that's the primary motivation 688 00:41:39.480 --> 00:41:44.320 of doing SSR. And what we've literally learned now is 689 00:41:44.400 --> 00:41:47.199 that you might be doing SSR with the goal of 690 00:41:47.320 --> 00:41:51.320 improving startup time and end up part of my French 691 00:41:52.000 --> 00:41:56.920 screwing yourself because your back end will not be able 692 00:41:56.960 --> 00:42:05.679 to properly handle the extra load. And I'm actually looking 693 00:42:05.880 --> 00:42:09.079 at the Crux data. We've spoken about it on the 694 00:42:09.079 --> 00:42:13.360 show on several occasions, so a quick reminder what it is. 695 00:42:14.559 --> 00:42:20.440 Google collects anonymous performance information from all browser sessions or 696 00:42:20.519 --> 00:42:23.719 Chrome browser sessions and unless you opt out, and it's 697 00:42:23.760 --> 00:42:25.880 not really easy to opt out, so assume it's more 698 00:42:26.000 --> 00:42:29.360 less everybody. And they are looking and then they are 699 00:42:29.400 --> 00:42:32.960 breaking it down by technologies used using the HTP archive. 700 00:42:33.440 --> 00:42:37.159 So basically for each domain, they're looking if a domain 701 00:42:37.239 --> 00:42:40.480 has sufficient amount of traffic, so they're looking at the 702 00:42:40.519 --> 00:42:43.639 top ten million domains more or less in terms of traffic. 703 00:42:44.239 --> 00:42:48.000 They're segmenting them by technologies used, and then they actually 704 00:42:48.400 --> 00:42:51.400 release this information to the world so you can actually 705 00:42:51.519 --> 00:42:56.280 see which technology is more likely to give you better performance. 706 00:42:56.840 --> 00:43:01.719 And I'm currently looking at the graph that compares all 707 00:43:01.760 --> 00:43:06.639 the websites within cracks, the WordPress websites, the Wix websites, 708 00:43:07.159 --> 00:43:11.039 and next JS websites. And I'm looking at what percentile 709 00:43:11.079 --> 00:43:15.239 of websites get good performance results. According to Google, it's 710 00:43:15.239 --> 00:43:20.119 the cob vitals. Okay, wow, yeah, And if we're talking 711 00:43:20.199 --> 00:43:25.119

about all technologies, we're talking about about fifty one percent, 712 00:43:25.199 --> 00:43:29.480 So approximately half of websites get good coed vitals and 713 00:43:29.559 --> 00:43:34.480 about half get poor web vitals. If we're talking about WordPress, 714 00:43:34.719 --> 00:43:37.880 the number is forty six percent. So WordPress is slightly 715 00:43:38.440 --> 00:43:42.800 lower than all technologies at large, but pretty close. So 716 00:43:43.000 --> 00:43:45.400 WordPress and all technologies is more or less the same. 717 00:43:45.800 --> 00:43:48.480 Maybe also the fact that like half the web runs 718 00:43:48.519 --> 00:43:53.199 on WordPress. Wis currently is at seventy three percent. 719 00:43:54.159 --> 00:43:56.880 And what is this website? Is the cracks data set? Right? 720 00:43:57.360 --> 00:43:59.320 Yes, I can send you. I'll put the link in 721 00:43:59.320 --> 00:43:59.920 the chat. 722 00:44:00.199 --> 00:44:02.119 An Houp going fantastic. I found it. 723 00:44:03.079 --> 00:44:05.840 So Wix has seventy three percent. So if you're building 724 00:44:05.880 --> 00:44:11.039 a website on Wix, there's a seventy three percent probability 725 00:44:11.400 --> 00:44:17.079 that you will be getting good performance. Okay, what's the number? Sorry, 726 00:44:17.159 --> 00:44:19.440 what's the number for next JS? Now you assume that 727 00:44:19.440 --> 00:44:22.159 next YS would be great because the whole thing about 728 00:44:22.199 --> 00:44:27.440 next GS is that it's as SARRD. So what's the 729 00:44:27.480 --> 00:44:28.920 probability for next gs? 730 00:44:28.960 --> 00:44:32.360 Can you guess feel like you're setting me up to 731 00:44:32.440 --> 00:44:37.039 let me down so well, I would have assumed before 732 00:44:37.079 --> 00:44:41.360 you said it, kind of skeptically that it'd be you know, yeah, 733 00:44:41.440 --> 00:44:44.519 somewhere around wis you know, sixty seventy percent, but you're 734 00:44:44.519 --> 00:44:47.119 making you're making it sound like it's probably like thirty percent. 735 00:44:47.679 --> 00:44:52.119 Exactly, it's twenty nine percent. So the probability of getting 736 00:44:52.119 --> 00:44:55.480 good performance when you're using NEXTJS, you know, you're you're 737 00:44:55.519 --> 00:44:58.960 putting all this effort into building let's say AI is 738 00:44:59.000 --> 00:45:01.760 not doing your entire job yet you're doing putting all 739 00:45:01.800 --> 00:45:05.719 this effort in encoding your website and react and hosting 740 00:45:05.800 --> 00:45:09.760 it and whatever you and your probability of getting good 741 00:45:09.800 --> 00:45:14.679 co ed vitals is less than thirty percent, And that's 742 00:45:15.039 --> 00:45:16.320 really unfortunate. 743 00:45:16.719 --> 00:45:20.679 Yeah, that's bad. So the stuff that Mateo is talking about, 744 00:45:20.760 --> 00:45:23.519 then do you do you get those kinds of performance 745 00:45:24.039 --> 00:45:27.239 increases kind of for free just by upgrading your note 746 00:45:27.320 --> 00:45:28.840 or make. 747 00:45:28.760 --> 00:45:31.840 It better by using what on top of next which 748 00:45:31.840 --> 00:45:32.719 it just works. 749 00:45:32.519 --> 00:45:36.400 So it's not and I can do it on versall, no. 750 00:45:36.559 --> 00:45:39.800 Versall ransom your own thing. So the fundamental problem, the 751 00:45:39.800 --> 00:45:42.920 fundamental point is next JS is designed to run well 752 00:45:42.960 --> 00:45:49.679 on versall and to run it well on your own infrastructure. 753 00:45:49.719 --> 00:45:52.320 You're missing a few components, and we added those few. 754 00:45:53.239 --> 00:45:55.599 So with verse cell out of the box, I'd be 755 00:45:55.679 --> 00:45:59.320 getting comparable performance to what I have. 756 00:46:00.079 --> 00:46:05.000 Absolutely no idea. I've not measured. I've not benchmarked their cloud. 757 00:46:05.199 --> 00:46:09.199 To be honest, I don't even it's a good product. 758 00:46:09.280 --> 00:46:12.440 I am a customer like Performati is a customer overrsell. Okay, 759 00:46:13.079 --> 00:46:15.119 we are as most startup. Is the greatest thing ever 760 00:46:15.199 --> 00:46:19.440 to be using perself. Okay, yeah, it's easy, it's easy. 761 00:46:19.559 --> 00:46:22.079 There are a lot of companies out there that cannot 762 00:46:22.199 --> 00:46:26.960 use yourself for all sorts of reasons, okay, from and 763 00:46:27.079 --> 00:46:28.280 that is critical. 764 00:46:29.000 --> 00:46:32.440 I know that the companies are running next gs on 765 00:46:32.599 --> 00:46:34.360 top of Amazon a. 766 00:46:34.480 --> 00:46:36.840 W Yeah exactly, So this is what I'm saying. 767 00:46:36.880 --> 00:46:40.440 Okay, they are running those things, and there is even 768 00:46:40.480 --> 00:46:42.480 a thing called open next run it on top of 769 00:46:42.760 --> 00:46:47.880 cloud Flare and other stuff. Okay, what we've done make 770 00:46:47.960 --> 00:46:51.039 it run very well on top of Kubernetus. 771 00:46:51.440 --> 00:46:55.239 By the way, I'm curious, what's your opinion of nest JS. 772 00:46:55.719 --> 00:47:00.239 You're asking me tough questions, then I am not I 773 00:47:00.320 --> 00:47:04.360 am I do have opinion. 774 00:47:04.440 --> 00:47:05.480 I have one opinion. 775 00:47:07.039 --> 00:47:09.880 Nest is the back end framework, specifically, isn't it? 776 00:47:10.360 --> 00:47:14.840 Nest is like NOE for Java developers I have. 777 00:47:17.840 --> 00:47:23.320 I have a problem with it. Okay, the problem I 778 00:47:23.400 --> 00:47:31.920 have with nest is its reliance on the old time 779 00:47:32.000 --> 00:47:35.880 skip decorators. So I don't know if you've been following 780 00:47:35.920 --> 00:47:39.119 the story of Java skip decorators with the year thirty nine. 781 00:47:39.239 --> 00:47:44.119 That standard that never ever materializes. It's been stuck in 782 00:47:44.760 --> 00:47:48.559 stage two and a half like forever

exactly. 783 00:47:49.800 --> 00:47:52.440 So the problem is that. 784 00:47:54.440 --> 00:47:57.480 Just you know what, before you do, let's just clarify 785 00:47:57.519 --> 00:48:00.719 that all our listeners I kind of assume that listeners 786 00:48:00.760 --> 00:48:05.760 were familiar with Kubernetes, which might have been either correct 787 00:48:05.880 --> 00:48:08.480 or incorrect. An assumption to make, but let's not assume 788 00:48:08.519 --> 00:48:13.679 the same for decorators. Decorators are a language feature in 789 00:48:13.800 --> 00:48:18.960 JavaScript or typescript which allow you to decorate function method 790 00:48:19.039 --> 00:48:23.400 classes and methods and properties. You put the ad sign 791 00:48:23.519 --> 00:48:26.599 and then some words, and basically it wraps the declaration 792 00:48:26.760 --> 00:48:30.239 in a function call and that function call can modify 793 00:48:31.239 --> 00:48:38.719 the actual class definition or the method definition. So for example, 794 00:48:38.800 --> 00:48:44.079 let's say the canonical example is I want to print 795 00:48:44.079 --> 00:48:47.719 out a log whenever a method enters and exits. Instead 796 00:48:47.760 --> 00:48:52.719 of explicitly adding a log everywhere, I can decorate the 797 00:48:52.800 --> 00:48:58.000 class maybe and thenates automatically. It basically modifies the behavior 798 00:48:58.079 --> 00:49:02.000 of all the class methods. Whenever I instantiate the instance, 799 00:49:02.039 --> 00:49:05.639 I get this functionality automatically. So it's a it's a 800 00:49:05.679 --> 00:49:08.920 way of implementing cross cutting concerns. 801 00:49:09.400 --> 00:49:13.559 Yes, it has two problems though, Okay. 802 00:49:13.840 --> 00:49:18.079 Oh, just to finish it's it's been a part it's 803 00:49:18.119 --> 00:49:24.679 it's very very popular in in Angular. Uh, it's been 804 00:49:24.800 --> 00:49:29.719 part of typescript for a while, and there's a proposal 805 00:49:29.800 --> 00:49:32.639 to add it to the JavaScript language itself, which, like 806 00:49:32.719 --> 00:49:34.280 we said, is kind of stuck in purgative. 807 00:49:35.159 --> 00:49:39.199 Yes, okay, let me explain the situation there. Okay, So 808 00:49:41.719 --> 00:49:45.039 the current so this pack. 809 00:49:46.639 --> 00:49:50.119 As it was originally proposed and done in typescript okay, 810 00:49:51.000 --> 00:49:56.920 all types types four and something okay, and before was 811 00:49:57.960 --> 00:50:01.159 was very hard to poly fil Okay. If you look 812 00:50:01.199 --> 00:50:06.320 at the JavaScript code that is being those syntax transpiles to, 813 00:50:07.960 --> 00:50:09.199 you can get very scared. 814 00:50:09.519 --> 00:50:15.480 Okay, lets let's go eight. It's like regenerator style things. 815 00:50:15.559 --> 00:50:18.320 I don't know if you ever looked at the cover 816 00:50:18.400 --> 00:50:22.639 of what our generator does, but is pretty ugly, okay, 817 00:50:22.760 --> 00:50:25.400 because there is a lot of global state. There is 818 00:50:25.480 --> 00:50:31.199 a lot of things happening. It's not nice. 819 00:50:31.760 --> 00:50:37.800 Then the standard changed in a way that it's essentially 820 00:50:37.880 --> 00:50:42.159 Java is essentially function wrappers okay, which can be attached 821 00:50:42.199 --> 00:50:46.760 to classes because classes are functions okay, or functions okay, 822 00:50:46.960 --> 00:50:52.039 which makes it way easier to to reason about this thing, okay. 823 00:50:52.159 --> 00:50:55.400 And also it's just a function, so it literally is 824 00:50:55.480 --> 00:50:59.679 easier to you know, implement some extent. 825 00:50:59.480 --> 00:51:02.719 The function that gets cold with the function as its 826 00:51:02.840 --> 00:51:05.679 argument and in context, and then it returns a function 827 00:51:05.800 --> 00:51:07.880 that gets used instead of the original function. 828 00:51:08.400 --> 00:51:10.800 Exactly, Very straightforward, Okay. 829 00:51:11.280 --> 00:51:16.880 I was just thinking, Yeah, it sounds like I'm still 830 00:51:16.920 --> 00:51:18.199 not sure I follow anyway. 831 00:51:18.719 --> 00:51:21.199 Anyway, It's it's simple, but there is a difference, okay. 832 00:51:21.280 --> 00:51:26.440 The old standard allowed to attach decorator to. 833 00:51:27.239 --> 00:51:29.760 Arguments parameters hmm. 834 00:51:29.960 --> 00:51:32.679 Yeah, and nest chess uses that a lot. 835 00:51:33.000 --> 00:51:39.159 Exactly, and in the new standard this is not possible. 836 00:51:38.719 --> 00:51:42.400 Anymore because they're not functions. They're just parameters. 837 00:51:42.000 --> 00:51:49.119 Exactly, and and that is my world problem. 838 00:51:49.639 --> 00:51:53.199 Okay, that nest chess is tied to a syntax which 839 00:51:53.239 --> 00:51:56.800 goes against the grain of where JavaScript is heading exactly. 840 00:51:57.840 --> 00:52:02.440 Also, its transpiled is really bad. That's it. 841 00:52:03.440 --> 00:52:05.920 I care a lot about the Java skity is being executed, 842 00:52:05.960 --> 00:52:10.239 and that is transpiled to some really two things that 843 00:52:10.280 --> 00:52:12.760 I would not have running on my servers. 844 00:52:16.119 --> 00:52:19.280 I'll say it again. It's really popular with people that 845 00:52:19.519 --> 00:52:24.079 come to note from Java, so I know a lot 846 00:52:24.119 --> 00:52:27.960 of back end developers that literally swear by it. It's 847 00:52:28.000 --> 00:52:30.119 a back end that you use as a back end 848 00:52:30.119 --> 00:52:33.519 for API. Course. Yes, it's it's not about necessa or 849 00:52:33.519 --> 00:52:35.599 anything like that exactly. 850 00:52:36.239 --> 00:52:37.519 But I am, I am. 851 00:52:37.679 --> 00:52:39.519 I have a side on all of this, so you 852 00:52:39.559 --> 00:52:41.039 can take all of what I said with a gain 853 00:52:41.119 --> 00:52:44.599 of salt. I have built to Fastify to solve this

problem. 854 00:52:44.719 --> 00:52:48.719 Been using it for the last decade and very happy 855 00:52:48.760 --> 00:52:49.039 with it. 856 00:52:49.599 --> 00:52:54.519 Yeah, it's really opinionated anyway. Moving But in terms of performance, 857 00:52:54.679 --> 00:52:56.760 what can you say aboutn SGS. 858 00:52:57.079 --> 00:53:01.519 I have not benchmarked it signific Okay. 859 00:53:01.920 --> 00:53:07.920 I suspect is slower than Core Express or Fastify. You 860 00:53:07.920 --> 00:53:10.920 can use Express or Fastify underneath, and I suspect is 861 00:53:11.119 --> 00:53:13.239 slower and it adds a little bit over it on top. 862 00:53:13.440 --> 00:53:15.960 Yeah, it has to kind of because it's literally built 863 00:53:16.000 --> 00:53:16.599 on top of it. 864 00:53:16.719 --> 00:53:25.559 So yes, I hope is as little as possible. But look, 865 00:53:25.679 --> 00:53:28.679 you're asking so you might get an article on it. 866 00:53:28.800 --> 00:53:31.760 So I'm my co found is telling me to do it, 867 00:53:31.880 --> 00:53:34.559 so to do the research. So I'll do the research. 868 00:53:34.760 --> 00:53:37.719 I can say that it's pretty popular. It's popular to 869 00:53:37.760 --> 00:53:41.360 the extent that it's boring, you know, like the best 870 00:53:41.360 --> 00:53:44.119 compliment you can give about the technology is that it's boring. 871 00:53:44.400 --> 00:53:46.400 Yeah, I know, I know, I do the same. 872 00:53:47.159 --> 00:53:50.400 Okay. So moving back to what we were talking about 873 00:53:50.440 --> 00:53:55.199 with so you talked about the surprising results that you got. Now, 874 00:53:55.840 --> 00:54:01.320 so obviously, if I'm running the bottom line is that 875 00:54:02.079 --> 00:54:07.400 if I'm running currently node on Amazon, and especially if 876 00:54:07.400 --> 00:54:09.679 I'm using it for a s SR, not necessarily, but 877 00:54:09.800 --> 00:54:12.760 especially if I'm using it for as SR, then I 878 00:54:12.840 --> 00:54:17.039 should be looking at what because bun won't save me 879 00:54:18.400 --> 00:54:22.639 pretty much. And okay, cool, anything else to add on 880 00:54:22.679 --> 00:54:23.719 that particular. 881 00:54:23.320 --> 00:54:26.119 Topic, Nope, I think I covered it all. 882 00:54:26.599 --> 00:54:29.639 So you had a list, let's go to the next one. Yay. 883 00:54:29.920 --> 00:54:32.920 So so just just on the issue of time. We 884 00:54:33.039 --> 00:54:38.199 have been recording for about an hour, which I'm totally 885 00:54:38.239 --> 00:54:42.519 fine sticking around for however long, but I need. 886 00:54:42.119 --> 00:54:46.039 To probably go at when these meeting ends that I 887 00:54:46.079 --> 00:54:47.320 have on my calendar. 888 00:54:47.119 --> 00:54:49.360 So that doesn't it just means that you want to 889 00:54:49.400 --> 00:54:51.320 bring it on and I bring you on again. 890 00:54:51.960 --> 00:54:58.519 Absolutely, I'm coming again. I'm coming again, of course. 891 00:54:58.719 --> 00:54:59.039 Okay. 892 00:54:59.119 --> 00:55:02.039 So the second one that I wanted to cover is 893 00:55:02.079 --> 00:55:05.360 the topic of flame graphs. Now, in all of these 894 00:55:05.400 --> 00:55:10.440 benchmarking to do a lot of optimizations and look at things, okay, 895 00:55:11.320 --> 00:55:16.639 and we created a brand new frame graph tool to 896 00:55:17.079 --> 00:55:18.440 do performance analysis. 897 00:55:18.960 --> 00:55:22.519 I don't know what a flame graph is, So imagine 898 00:55:22.559 --> 00:55:23.840 that you are. 899 00:55:25.280 --> 00:55:27.679 You want to do a performance optimization. Okay, you have 900 00:55:27.880 --> 00:55:30.239 you know your application is low, and you want to 901 00:55:30.239 --> 00:55:31.000 know where it's low. 902 00:55:31.440 --> 00:55:35.280 Okay. A flame graph shows. 903 00:55:34.880 --> 00:55:40.440 You the hottest points in your code base that consume 904 00:55:40.599 --> 00:55:41.239 the most time. 905 00:55:42.400 --> 00:55:44.239 Okay, I've seen them, I just didn't know that's what 906 00:55:44.280 --> 00:55:45.800 they were called exactly. 907 00:55:45.960 --> 00:55:46.239 Okay. 908 00:55:46.280 --> 00:55:50.079 So Cinema GRAPHA and a new relic, so yeah. 909 00:55:49.880 --> 00:55:50.199 They can. 910 00:55:50.280 --> 00:55:52.519 You can click a button and get the flame graph. 911 00:55:52.599 --> 00:55:56.000 The key point is you the flame graph abstract time. 912 00:55:56.039 --> 00:55:58.800 So it takes an internet of time and says, let 913 00:55:58.840 --> 00:56:01.400 me represent this the function calls that happen in this 914 00:56:01.519 --> 00:56:05.559 interval of time. Okay, because in that way, I know 915 00:56:05.679 --> 00:56:08.239 how much a function is being called compared to the others. 916 00:56:08.280 --> 00:56:09.559 So I can make a comparison. 917 00:56:10.000 --> 00:56:14.639 Okay, so I can, so we can detect how much 918 00:56:14.719 --> 00:56:16.559 function was called and how hot it is. 919 00:56:16.920 --> 00:56:21.719 Okay, but I've been doing this for for a long time. 920 00:56:21.800 --> 00:56:25.000 I created the second kind of two third generation tool 921 00:56:25.039 --> 00:56:28.119 that I create with these kind of things. Okay, a 922 00:56:28.280 --> 00:56:31.679 new version of note require differently tools to get it done. 923 00:56:32.239 --> 00:56:34.559 PingER Cross is the last, but probably there will be 924 00:56:34.599 --> 00:56:35.639 another one in a few years. 925 00:56:35.800 --> 00:56:36.079 Okay. 926 00:56:36.519 --> 00:56:40.880 Now, somebody who's listening to us and here's a flame 927 00:56:40.960 --> 00:56:46.639 graft to measure NOTE performance might be thinking, Hey, why 928 00:56:46.679 --> 00:56:49.480 do I need this from a platformatic where I have 929 00:56:49.559 --> 00:56:57.239 it built into dev tools for like forever for why 930 00:56:57.280 --> 00:56:59.199 do I need at another one? What's wrong

with the 931 00:56:59.199 --> 00:57:00.840 one that's built in death tools. 932 00:57:01.280 --> 00:57:08.239 The tools gives you a flame shart, not a flame draft. 933 00:57:09.599 --> 00:57:15.840 It's not as readable, it doesn't give you the full picture. Also, 934 00:57:17.519 --> 00:57:23.000 it's the way. The part of the problem is the 935 00:57:23.039 --> 00:57:27.840 collection of that data is very expensive, so by running 936 00:57:28.000 --> 00:57:34.039 it in the tools, you are essentially altering, like the 937 00:57:34.559 --> 00:57:39.079 act of observing something modifies the behavior. 938 00:57:40.440 --> 00:57:44.119 Quantum mechanics, right, Yeah, it's quantum mechanics at play. 939 00:57:44.400 --> 00:57:47.719 Yes, just to clarify, I'm aware of the issues and 940 00:57:47.760 --> 00:57:50.039 I totally agree with you. I'm just bringing it up 941 00:57:50.079 --> 00:57:52.400 so you can rebut it perfect. 942 00:57:52.480 --> 00:57:52.760 Okay. 943 00:57:52.800 --> 00:57:55.639 So it's quantum mechanics at work. You observe it, you 944 00:57:55.760 --> 00:57:56.199 change it. 945 00:57:56.400 --> 00:57:56.760 Okay. 946 00:57:57.440 --> 00:58:01.239 The way the what we have done instead of relying 947 00:58:01.320 --> 00:58:04.800 on the inspector to to get that data, okay, we 948 00:58:04.840 --> 00:58:08.599 are using a C plus plus module from our good 949 00:58:08.639 --> 00:58:12.039 friends at Data Dog that they're using their agent to 950 00:58:12.079 --> 00:58:14.920 collect those data from C plus plus straight out of 951 00:58:15.000 --> 00:58:19.039 the eight in the fastest possible way. So it does 952 00:58:19.119 --> 00:58:24.199 not interfere much with the application itself to the point 953 00:58:24.239 --> 00:58:26.760 that you can use it in production to collect those 954 00:58:26.760 --> 00:58:29.679 flame graphs. Now, if you try to enable the inspector, 955 00:58:29.760 --> 00:58:32.079 you are looking at a fifty percent performance job. 956 00:58:32.960 --> 00:58:35.920 It's beyond that. First of all, let's start with the 957 00:58:35.960 --> 00:58:39.280 fact that you probably don't want to attach the development 958 00:58:39.360 --> 00:58:44.000 tool to your production environment. Yes, somebody might click pause, 959 00:58:49.840 --> 00:58:55.519 Oh yeah, you're not going to be running in production 960 00:58:55.719 --> 00:59:02.920 Node servant production with as inspect you know, yeah, yeah, 961 00:59:02.960 --> 00:59:05.719 and you're not and you're not going to believe that 962 00:59:05.719 --> 00:59:08.280 that port open. Well, you know, you can do it 963 00:59:08.360 --> 00:59:14.000 as such tunneling and whatnot. But anyway, and like Matteo said, 964 00:59:14.239 --> 00:59:22.639 the overhead of actually doing a flame chart is really high. 965 00:59:21.960 --> 00:59:25.760 And by the way, also for that same reason, it's 966 00:59:25.800 --> 00:59:29.119 time constrained. You can't run it forever. You can run 967 00:59:29.199 --> 00:59:32.119 it for a very limited amount of time. Now, I 968 00:59:32.559 --> 00:59:37.559 have in the past implemented certain scenarios in certain applications 969 00:59:37.599 --> 00:59:42.039 where we sent a signal or an event into a 970 00:59:42.079 --> 00:59:47.960 process to kind of trigger it. But it's you can again, 971 00:59:48.000 --> 00:59:50.159 you can only run these sort of things for a 972 00:59:50.199 --> 00:59:51.519 really limited amount of time. 973 00:59:52.639 --> 00:59:53.119 Exactly. 974 00:59:54.320 --> 00:59:56.599 You can also still run this kind of tool for 975 00:59:56.639 --> 01:00:01.079 a limited amount of time, So it's a you know, 976 01:00:01.199 --> 01:00:03.840 you can even the way you do it, you run 977 01:00:03.920 --> 01:00:08.679 it for essentially one minute to collect what's happening. And 978 01:00:09.400 --> 01:00:14.400 in our platform, we can even start collecting it when 979 01:00:14.519 --> 01:00:18.400 usage goes over a certain threshold, so that you know, 980 01:00:18.519 --> 01:00:24.119 you don't you're not wasting money. But it's actually very good, 981 01:00:24.360 --> 01:00:26.800 so you can actually. 982 01:00:28.159 --> 01:00:31.719 Even running in production get good data out. Data Doctor 983 01:00:31.760 --> 01:00:33.920 is running this in production all the time. It's fantastic. 984 01:00:34.679 --> 01:00:37.039 By the way, I have to say, it was really funny. 985 01:00:37.079 --> 01:00:39.679 So when I found out about this tool and I 986 01:00:39.719 --> 01:00:42.960 started using it, and I started experimenting with it, and 987 01:00:43.000 --> 01:00:46.280 I reported like six issues or something like that back 988 01:00:46.320 --> 01:00:49.599 to Mateo and the guys and the gang, and like 989 01:00:49.679 --> 01:00:53.119 within one day it fixed all of them. It was 990 01:00:53.840 --> 01:00:54.800 it was really amusing. 991 01:00:55.159 --> 01:01:00.320 Yeah, very fast. It's so there is that. Okay, tool 992 01:01:00.480 --> 01:01:04.000 was great. Last week. I think we shaped something new. 993 01:01:04.239 --> 01:01:10.119 We shipped a way to get the representation of the 994 01:01:10.159 --> 01:01:13.960 flamegraph as marked down, so you can essentially feed it 995 01:01:14.079 --> 01:01:17.639 to an l LM and it will just fix things. 996 01:01:18.800 --> 01:01:24.519 So it will basically understand where your bottlenecks are in 997 01:01:24.599 --> 01:01:29.800 terms of performance and focus on them automatic exactly. Like 998 01:01:29.960 --> 01:01:34.400 this function is slow, but it's not just slow. It's impacting. 999 01:01:35.199 --> 01:01:38.000 It's being called a lot, so it has a significant impact. 1000 01:01:38.199 --> 01:01:40.840 Or that function is hardly used, there's no point in 1001 01:01:40.840 --> 01:01:42.039 optimizing exactly. 1002 01:01:43.440 --> 01:01:46.559 Yeah, but like you said, Matteo, it seems also that 1003 01:01:47.519 -->

01:01:51.639 some of the fixes are pretty commonly understood and easy 1004 01:01:51.679 --> 01:01:54.039 for the LLM to implement, and so I can also 1005 01:01:54.480 --> 01:01:57.760 you know, it's like add add a database index or hey, 1006 01:01:57.920 --> 01:01:59.599 if you can't if you do it this way instead 1007 01:01:59.639 --> 01:02:02.000 of that way. This structure is more performant than that one. 1008 01:02:02.079 --> 01:02:06.320 And you know it's it's it's little changes, but they're 1009 01:02:07.320 --> 01:02:10.599 low risk, low risk, and the LAM is great at it. 1010 01:02:10.800 --> 01:02:14.840 Okay, I called the low hanging fruits. Okay, so the 1011 01:02:14.880 --> 01:02:18.119 low hanging fruits are super easy to do. 1012 01:02:18.360 --> 01:02:19.559 For the alarms. Okay. 1013 01:02:20.119 --> 01:02:23.800 Now from the frame graph you can even spot some 1014 01:02:23.960 --> 01:02:28.679 bad architectural patterns happening. And the bad architectural patterns are 1015 01:02:28.800 --> 01:02:32.079 harder so to figure out. 1016 01:02:32.280 --> 01:02:36.960 So during my research I've done, I opened a few bugs. Okay. 1017 01:02:37.800 --> 01:02:42.119 One is on a React router. I found that it 1018 01:02:43.119 --> 01:02:47.119 for every rendering its schedules a timer that's never cleared, 1019 01:02:47.719 --> 01:02:51.400 so it runs to completion. It's it's like five second, 1020 01:02:52.159 --> 01:02:55.840 but that cause it's becomes a hot spot for the 1021 01:02:55.840 --> 01:02:56.639 garbage collection. 1022 01:02:57.519 --> 01:02:59.679 Right. I was going to say, it's a memory leak, 1023 01:02:59.719 --> 01:03:01.920 but it's nice. It does get it. 1024 01:03:04.400 --> 01:03:08.239 Eventually, yes, but it's enough that it gets by. It'll 1025 01:03:08.400 --> 01:03:11.559 keep memory located for more and it's enough for that 1026 01:03:11.719 --> 01:03:15.119 memory to be moved to old space. So it's actually 1027 01:03:15.159 --> 01:03:19.159 creating a some problems for the wrong time that's actually 1028 01:03:19.199 --> 01:03:23.039 not needed. That could avoid could be avoided. Okay, I 1029 01:03:23.039 --> 01:03:26.480 sent a patch. Another one that I found recently is 1030 01:03:26.639 --> 01:03:30.159 a bug on with this tool is on GT. 1031 01:03:30.480 --> 01:03:31.519 I don't know if you know GT. 1032 01:03:32.199 --> 01:03:34.119 By the way, I found this a very horrible bug 1033 01:03:34.239 --> 01:03:36.079 and I am really bad. 1034 01:03:36.480 --> 01:03:38.639 Let me pass it to you because. 1035 01:03:40.559 --> 01:03:44.079 And GT is a module a lot than sixty million 1036 01:03:44.119 --> 01:03:45.400 times per week something like that. 1037 01:03:46.280 --> 01:03:46.599 GT. 1038 01:03:47.440 --> 01:03:51.199 Yeah, GT, this is a bug. I passed the bag 1039 01:03:51.199 --> 01:03:53.639 in the chat J I T I yeah. 1040 01:03:54.440 --> 01:03:58.440 It makes any any function that export in a function 1041 01:03:59.199 --> 01:04:01.000 between sixteen seventy times lower. 1042 01:04:01.519 --> 01:04:02.320 What is chitty. 1043 01:04:03.400 --> 01:04:08.400 It's a module to load the typescript straight from node, 1044 01:04:08.599 --> 01:04:12.920 similar to tes what tes no does, but that differently 1045 01:04:13.039 --> 01:04:18.159 typed syme of things is used by yes, lint, nxt, netro, 1046 01:04:18.960 --> 01:04:19.800 post CSS. 1047 01:04:22.480 --> 01:04:22.920 I don't know. 1048 01:04:23.559 --> 01:04:26.559 Tailwind is very popular in front and I have I 1049 01:04:26.559 --> 01:04:28.800 had no idea this even existed until it showed up 1050 01:04:28.800 --> 01:04:29.800 in my flamegraphs. 1051 01:04:30.239 --> 01:04:33.840 Man, don't you love it when people use proxies? 1052 01:04:34.360 --> 01:04:37.719 Yes? You see see I have you know it's it's 1053 01:04:37.760 --> 01:04:40.960 it's one of the most misused technologies that you can 1054 01:04:41.000 --> 01:04:43.440 get in that Like, if you're using a proxy, you 1055 01:04:43.480 --> 01:04:46.239 should ask am I doing it wrong? 1056 01:04:47.480 --> 01:04:51.639 And the answer would probably be yes, yes, not always, 1057 01:04:51.719 --> 01:04:53.719 not one hundred percent of the time, but you should 1058 01:04:53.760 --> 01:04:55.679 definitely be able to justify yourself. 1059 01:04:56.039 --> 01:04:59.199 Yeah. So I'm going to kind of push us to 1060 01:04:59.239 --> 01:05:02.079 wrap up this particular topic because we only have like 1061 01:05:02.119 --> 01:05:03.079 eight minutes left. 1062 01:05:03.079 --> 01:05:06.679 So I just have to say that before we conclude 1063 01:05:06.679 --> 01:05:12.840 on this, that the ability to gather performance information throughout 1064 01:05:12.880 --> 01:05:19.159 the lifetime of a process in production is invaluable. You know, 1065 01:05:19.280 --> 01:05:24.960 we've had similar technologies in other programming languages, like the 1066 01:05:25.280 --> 01:05:32.280 what's it called in Java? Java has the Java black 1067 01:05:32.400 --> 01:05:37.440 box sort of thing, I forget the name. That they 1068 01:05:37.960 --> 01:05:42.000 have the ability to collect performance information without usually without 1069 01:05:42.039 --> 01:05:46.079 significantly impacting runtime performance, that you can do it in production. 1070 01:05:46.199 --> 01:05:48.840 The ability to do the same thing or similar thing 1071 01:05:49.480 --> 01:05:54.679 in for node is really invaluable. And I'm really grateful 1072 01:05:54.719 --> 01:05:58.119 for Mattel and everybody Platformatic for releasing this game. 1073 01:05:59.400 --> 01:06:02.599 Is there free tried out if you want, if you need, 1074 01:06:02.880 --> 01:06:06.840 you know, talk a little bit about my company enough. 1075 01:06:07.039 --> 01:06:11.840 But if you using this tag and you it's core 1076 01:06:11.880 --> 01:06:13.920 part of what you do, you can reach out. 1077 01:06:14.000 -->

01:06:15.559 We'll be very happy to help you. 1078 01:06:18.159 --> 01:06:22.480 Cool. All right, Well, I'm going to push us two picks. Uh, 1079 01:06:22.840 --> 01:06:24.679 this is where we do shout outs about stuff we 1080 01:06:24.840 --> 01:06:30.000 like and uh yeah, I know that you're kind of 1081 01:06:30.079 --> 01:06:32.599 under a time crunch Matereo, do you want to go first? 1082 01:06:32.639 --> 01:06:34.119 Or do you want one of us to go first? 1083 01:06:34.559 --> 01:06:34.800 Oh? 1084 01:06:35.719 --> 01:06:38.599 Let me let you go first and then I go last. 1085 01:06:38.960 --> 01:06:41.159 Okay, Dan, what are your picks? 1086 01:06:41.599 --> 01:06:46.679 Okay? So I've got two picks. The first pick is. 1087 01:06:46.760 --> 01:06:52.559 We've been playing this amusing game in our family these 1088 01:06:52.599 --> 01:06:57.320 past couple of weeks. It's called Hitster. It's this kind 1089 01:06:57.360 --> 01:07:00.719 of a game where you've got a certain like deck 1090 01:07:00.760 --> 01:07:05.960 of cards and each card has this QR code that 1091 01:07:06.039 --> 01:07:10.400 you scan, and it goes with your phone and it 1092 01:07:10.519 --> 01:07:13.960 goes to the and it actually starts playing that song. 1093 01:07:15.519 --> 01:07:18.880 I think it uses Spotify, but I'm not. I don't 1094 01:07:18.880 --> 01:07:21.239 remember off the top of my head. And the idea 1095 01:07:21.480 --> 01:07:24.159 is that you play every you divide into teams. You 1096 01:07:24.199 --> 01:07:28.320 play the song for about twenty seconds, and then the 1097 01:07:28.400 --> 01:07:33.719 team that turn it currently is needs to guess the artist, 1098 01:07:34.360 --> 01:07:36.519 the title of the song, the artist, and the year 1099 01:07:36.559 --> 01:07:39.920 in which it came out. If you need to guess 1100 01:07:39.960 --> 01:07:42.599 at least two to get a point, If you guess 1101 01:07:42.639 --> 01:07:46.639 all three, you get two points. Something along these lines. 1102 01:07:47.639 --> 01:07:50.119 And we've been playing this in our family, and it's 1103 01:07:50.639 --> 01:07:53.679 songs from all eras. In the case of Israel, it's 1104 01:07:53.679 --> 01:07:58.480 both Israeli music and you know international or American music 1105 01:07:58.559 --> 01:08:01.800 or British music, and it's really a lot of fun. 1106 01:08:01.880 --> 01:08:06.239 We enjoy it very very much. You know, you listen 1107 01:08:06.280 --> 01:08:09.800 to music, what's you know and and try to guess 1108 01:08:10.320 --> 01:08:12.960 the songs. It's it's a lot of fun. So that 1109 01:08:13.000 --> 01:08:17.359 would be my first pick, and the second pick is 1110 01:08:17.399 --> 01:08:21.720 again is actually Matteo. I want to mention that on 1111 01:08:22.439 --> 01:08:27.760 X Matteo, you put out videos on a fairly regular basis. 1112 01:08:28.279 --> 01:08:31.279 It's they're relatively short usually I think they're like twenty 1113 01:08:31.319 --> 01:08:35.800

minutes each, usually about a variety of topics related to 1114 01:08:36.000 --> 01:08:41.520 know the Jass development obviously, and they're uniformly excellent, and 1115 01:08:41.600 --> 01:08:45.359 I highly recommend following Matteo and watching the videos that 1116 01:08:45.880 --> 01:08:50.199 come out. So that would be my second pick, And 1117 01:08:50.359 --> 01:08:51.600 those are my picks for today. 1118 01:08:52.159 --> 01:08:55.680 Awesome, Steve, what are your picks. 1119 01:08:55.600 --> 01:09:00.439 Before I get to the high point of every episode, 1120 01:09:00.479 --> 01:09:02.920 which are the dad jokes of the week. An interesting 1121 01:09:03.279 --> 01:09:08.920 video that popped up on YouTube last night by Jeffrey Ray. 1122 01:09:09.199 --> 01:09:13.800 Jeffrey Ray, who is the owner creator of Larra Cass, which, 1123 01:09:13.800 --> 01:09:16.560 if not Larabelle Community is a pretty well known training 1124 01:09:16.600 --> 01:09:21.159 platform for Larabell and on many many other topics as well. 1125 01:09:21.680 --> 01:09:24.560 And he titled it I'm Done, and it's just sort 1126 01:09:24.560 --> 01:09:29.479 of a a it's about thirteen fourteen minutes where he's 1127 01:09:29.520 --> 01:09:31.960 just talking about the impact that AI has had on 1128 01:09:32.199 --> 01:09:34.439 Larrakass and they just had to layoff a bunch of people, 1129 01:09:35.399 --> 01:09:39.600 similar to what happened with Tailwind and the you know, 1130 01:09:39.880 --> 01:09:44.399 sort of the use of AI and coding and you know, 1131 01:09:44.439 --> 01:09:49.600 the pros and cons and sort of almost having to 1132 01:09:49.880 --> 01:09:54.760 use it or get left behind. He's a I gotta 1133 01:09:54.800 --> 01:09:56.119 agree with him in a lot of what he says. 1134 01:09:56.119 --> 01:09:58.600 It's only about fourteen minutes, so it's it's a good watch, 1135 01:09:58.680 --> 01:10:03.159 but it's I think it addresses the reality of AI 1136 01:10:03.319 --> 01:10:06.520 encoding and how it's impact impacting things both I think, 1137 01:10:06.560 --> 01:10:08.079 I think both for good and bad. 1138 01:10:08.680 --> 01:10:12.800 I think we should probably bring a bunch of open 1139 01:10:12.840 --> 01:10:19.920 source creators, people like Matteo, like maybe uh, we've had 1140 01:10:19.920 --> 01:10:24.079 a of the Joel we we've had on the show 1141 01:10:25.720 --> 01:10:29.479 talking about the slint and t slint uh and and 1142 01:10:29.640 --> 01:10:34.760 basically talk about the impact that AI is having on 1143 01:10:34.760 --> 01:10:39.119 on open source developers. It's really cutting the branch that 1144 01:10:39.239 --> 01:10:41.840 a lot of them are sitting on financially. 1145 01:10:41.720 --> 01:10:43.960 On that upper Ever note, I'll get to the dad 1146 01:10:44.039 --> 01:10:49.239 jokes of the week. So, first of all, pretty straightforward 1147 01:10:49.279 --> 01:10:53.199 question here, what do you call a pony with a 1148 01:10:53.279 --> 01:10:56.600 sore throat? A little horse? 1149 01:10:59.439 --> 01:10:59.600 Right? 1150 01:11:02.479 --> 01:11:03.720 So good, thank you, Mateo. 1151 01:11:03.800 --> 01:11:04.159 I love that. 1152 01:11:07.039 --> 01:11:09.319 So everybody knows who Alan Turing is, right, the term 1153 01:11:09.399 --> 01:11:14.000 touring complete Uh, you know in computer science Enigma and yeah, 1154 01:11:14.039 --> 01:11:16.760 he cracked the Enigma codes in World War Two, but 1155 01:11:17.800 --> 01:11:20.840 nobody knows his sister Kay, who provided drink, snacks and 1156 01:11:20.840 --> 01:11:26.560 sandwiches for him and his colleagues. I was in catering, Yeah, 1157 01:11:27.000 --> 01:11:27.319 right right. 1158 01:11:27.800 --> 01:11:29.319 It took me a minute. Yeah. 1159 01:11:29.359 --> 01:11:33.920 And then finally, uh, why does Spider Man hate driving 1160 01:11:33.960 --> 01:11:41.640 with his evil twin because he's a bad parallel parker? Oh? 1161 01:11:41.960 --> 01:11:43.760 Thank you. Those are the dad jokes of the week. 1162 01:11:45.000 --> 01:11:47.960 I usually don't laugh at the dad jokes, but Mateo's 1163 01:11:48.039 --> 01:11:49.520 reaction was priceless. 1164 01:11:50.000 --> 01:11:51.279 That's half the fun sometimes. 1165 01:11:52.000 --> 01:11:54.720 Oh all right, I'm gonna jump in with some picks. 1166 01:11:55.000 --> 01:11:56.960 The first one I always do a board game pick, 1167 01:11:58.439 --> 01:12:02.239 and this one we got for Christmas. Man, has it 1168 01:12:02.279 --> 01:12:04.199 been that long since I've been able to get on 1169 01:12:04.279 --> 01:12:10.960 and record. Yes, thanks Steve. He made me feel better. Anyway. 1170 01:12:11.600 --> 01:12:13.359 The game that one of the games we got was 1171 01:12:13.359 --> 01:12:18.119 Everdell came out twenty eighteen. My wife and I played it, 1172 01:12:18.119 --> 01:12:20.680 so it was just two players with us. It probably 1173 01:12:20.680 --> 01:12:22.880 took us an hour, maybe a little longer to play. 1174 01:12:23.399 --> 01:12:25.920 We were learning it as we went though. I think 1175 01:12:25.960 --> 01:12:29.239 I played it with my sister and my wife another 1176 01:12:29.319 --> 01:12:31.760 time and it took us about the same amount of 1177 01:12:31.760 --> 01:12:33.560 time because at that point we knew what we were doing. 1178 01:12:34.439 --> 01:12:38.800 And anyway, what it is, it's actually the board's kind 1179 01:12:38.840 --> 01:12:42.560 of cool because it's got this tree that you kind 1180 01:12:42.560 --> 01:12:46.359 of slide together. It's made out of cardboard, but it 1181 01:12:46.399 --> 01:12:48.840 looks really cool when it's set up and it's sitting 1182 01:12:48.880 --> 01:12:51.720 on the board. Reminds me a little bit of the 1183 01:12:52.439 --> 01:12:59.880 Tower in Fate of the Fellowship, which is a pandemic 1184 01:13:00.279 --> 01:13:05.079 like game based on Lord of the Rings. But anyway, 1185 01:13:05.159 -->

01:13:09.319 so Everdell plays up to four players and you're trying 1186 01:13:09.319 --> 01:13:11.960 to collect resources in order to fill missions, and you're 1187 01:13:11.960 --> 01:13:18.520 building a little town of animals in you know, basically 1188 01:13:18.600 --> 01:13:20.600 in front of you, and that gives you special ability. 1189 01:13:20.680 --> 01:13:25.279 So it's a pretty standard kind of game. It's just 1190 01:13:25.399 --> 01:13:29.000 you know, a couple of little nuances and stuff. Anyway, 1191 01:13:29.399 --> 01:13:32.159 it was fun. Definitely enjoyed it. It's not. 1192 01:13:33.640 --> 01:13:34.359 It's not so. 1193 01:13:34.560 --> 01:13:37.479 Novel or different or whatever that I would just go 1194 01:13:37.560 --> 01:13:41.880 and reach for it on my own on a regular basis. 1195 01:13:42.039 --> 01:13:44.319 But I did enjoy it. It's definitely a game worth, 1196 01:13:44.520 --> 01:13:47.439 you know, having him playing, and the artwork on it 1197 01:13:47.479 --> 01:13:52.640 is incredible. So I'm going to pick Everdell board game 1198 01:13:52.680 --> 01:13:58.119 has a board game weight of two point eighty three, 1199 01:13:58.239 --> 01:14:02.439 and that means that it's it's it's a fairly involved game, 1200 01:14:02.479 --> 01:14:04.439 but if you've played a lot of the other resource 1201 01:14:04.479 --> 01:14:11.119 gathering deck building kind of games, it's nothing outside of 1202 01:14:11.159 --> 01:14:16.800 the realm of what you're what you've probably already played. 1203 01:14:17.199 --> 01:14:24.960 So I'll pick that, and then what else. I started 1204 01:14:25.000 --> 01:14:28.199 watching land Man and I've been enjoying that. That's on 1205 01:14:28.239 --> 01:14:31.239 Paramount Plus, and so I'm going to pick that. And 1206 01:14:31.279 --> 01:14:34.159 then I watched My wife and I started watching the show, 1207 01:14:34.239 --> 01:14:36.800 and she's a little bit sensitive to people dropping f 1208 01:14:36.880 --> 01:14:39.680 bombs during shows, and so if there's too much of that, 1209 01:14:39.720 --> 01:14:41.720 she won't watch it. And so we watched like the 1210 01:14:41.760 --> 01:14:45.520 first episode way back when of eleven twenty two sixty three, 1211 01:14:47.399 --> 01:14:51.880 which is, uh, there's this history teacher. His friend's been 1212 01:14:51.920 --> 01:14:54.800 going back in time to try and save JFK from 1213 01:14:54.840 --> 01:14:59.279 being assassinated. And so anyway, he takes all of his 1214 01:14:59.319 --> 01:15:03.000 friends research and you know, all the preparation that he's done, 1215 01:15:03.000 --> 01:15:06.600 and he goes back to save JFK and right then 1216 01:15:06.640 --> 01:15:08.399 come back to the future where it's supposed to be 1217 01:15:08.399 --> 01:15:11.479 better because you know, Linda Johnson wouldn't have gotten us 1218 01:15:11.479 --> 01:15:14.039 into the Vietnam War and a whole bunch of other 1219 01:15:14.439 --> 01:15:16.840 things that other people did that you know, this particular 1220 01:15:16.880 --> 01:15:20.640 guy didn't like wouldn't have happened. And anyway, so I'm 1221 01:15:20.640 --> 01:15:23.000 not going to spoil how it changes the future, even 1222 01:15:23.000 --> 01:15:26.359 though the show's like ten years old, just in case 1223 01:15:26.359 --> 01:15:29.560 you want to go watch it, because honestly, it's it's 1224 01:15:29.600 --> 01:15:34.720 actually pretty like the whole journey's pretty pretty interesting to 1225 01:15:34.720 --> 01:15:38.399 watch him go through and you know, encounter people from 1226 01:15:38.399 --> 01:15:41.680 the past and stuff like that. So I'm going to pick. 1227 01:15:42.119 --> 01:15:44.279 So you've got to pick Prime Month plus. How are 1228 01:15:44.319 --> 01:15:46.479 you enjoying Star Trek Starfleet Academy. 1229 01:15:47.039 --> 01:15:51.239 I have not watched it at all. I've seen ads 1230 01:15:51.239 --> 01:15:53.119 for it, but I haven't watched it. Is it good. 1231 01:15:53.840 --> 01:15:57.720 I've not watched it. I'm going based on the reactions 1232 01:15:57.800 --> 01:16:02.960 and the critics. You know, the story that they released 1233 01:16:02.960 --> 01:16:08.279 the first episode for free on YouTube and a nerdratic 1234 01:16:09.359 --> 01:16:12.319 in competition with him basically just put a video of 1235 01:16:13.000 --> 01:16:15.760 doll of Spock sitting on a chair and he got 1236 01:16:15.800 --> 01:16:16.520 more views. 1237 01:16:16.800 --> 01:16:19.520 Oh funny. I have to say that. 1238 01:16:20.960 --> 01:16:23.680 I understand that they did. It's Star Treks Acolyte. 1239 01:16:24.359 --> 01:16:29.680 Okay, Yeah, I've I've liked a lot of stuff on 1240 01:16:30.600 --> 01:16:34.319 Paramount Plus, but yeah, that's not That's not one that 1241 01:16:34.359 --> 01:16:40.000 I've been watching. We liked Star Trek Discovery and Star 1242 01:16:40.079 --> 01:16:44.960 Trek Piccard, but anyway, I haven't watched that one yet. 1243 01:16:45.600 --> 01:16:50.159 So yeah, so Landman and eleven sixty three, which I 1244 01:16:50.159 --> 01:16:54.560 think was originally on Prime and is now on Netflix. Anyway, Tayo, 1245 01:16:54.600 --> 01:16:55.520 what are your picks? 1246 01:16:56.319 --> 01:16:57.960 So the first one is. 1247 01:16:59.399 --> 01:17:02.640 I've been taking a lot with Ai thinks, Okay, as 1248 01:17:02.680 --> 01:17:07.439 everybody and I have having a lot of fun these 1249 01:17:07.520 --> 01:17:13.680 last few days on a project called PI. It's a 1250 01:17:13.720 --> 01:17:17.720 little open source codeing agent that is highly customizable. The 1251 01:17:17.880 --> 01:17:22.000 heart of malt Bot now cloud bot whatever is what 1252 01:17:22.039 --> 01:17:26.239 that thing uses internally to do work. Okay, and it's 1253 01:17:26.279 --> 01:17:30.000 super fun to use and very fun to modify and 1254 01:17:30.479 --> 01:17:31.960 change things and so on and so forth. 1255

01:17:32.680 --> 01:17:36.079 Is really fun. So I've been tinkling with that. 1256 01:17:37.239 --> 01:17:39.720 Also, I can use it with Ulama and even have 1257 01:17:39.760 --> 01:17:44.079 a local model running very easily, and I add some success, 1258 01:17:44.119 --> 01:17:46.319 some success, so I can run the full AI think 1259 01:17:46.359 --> 01:17:50.279 on my machine. Not that I want to, but it's possible. 1260 01:17:50.279 --> 01:17:52.880 So I can literally use those things on a plane 1261 01:17:52.880 --> 01:17:56.079 if I want to, which is something that I really 1262 01:17:56.119 --> 01:17:57.039 wanted to achieve. 1263 01:17:57.520 --> 01:18:04.239 So I have done had fun with that. Okay, yeah, 1264 01:18:03.800 --> 01:18:07.239 it's very easy. The other thing it's we're talking with 1265 01:18:07.279 --> 01:18:10.039 the eye is tile scale. If you're not using tilescale, 1266 01:18:10.319 --> 01:18:15.039 tailscale is you're missing out. Okay. You can interconnect all 1267 01:18:15.039 --> 01:18:17.119 your devices, servers and so and so force in a 1268 01:18:17.159 --> 01:18:23.159 single network. It's phenomenal. Okay, I'm a fan. So try 1269 01:18:23.199 --> 01:18:26.239 that and combination with those two together, I can more 1270 01:18:26.319 --> 01:18:29.119 or ize call things on the phone if I want to, 1271 01:18:29.720 --> 01:18:31.880 and whenever I go, I can just have you know, 1272 01:18:31.960 --> 01:18:36.239 checking on my agents doing things. And yeah, I'm becoming 1273 01:18:36.239 --> 01:18:37.640 one of those people apparently. 1274 01:18:38.319 --> 01:18:42.720 So the problem that I have right now, I know 1275 01:18:42.840 --> 01:18:45.039 that a lot of people are running multiple agents at 1276 01:18:45.079 --> 01:18:49.439 the same time. I'm having problems with that multitasking between 1277 01:18:49.800 --> 01:18:54.840 different tasks myself, so it's difficult for me to schedule 1278 01:18:55.039 --> 01:18:57.720 enough agents to do things all the time in the background. 1279 01:18:59.600 --> 01:19:03.239 I maybe it's a fault of my own as a 1280 01:19:03.279 --> 01:19:08.319 manager or something, but the end result is that I 1281 01:19:08.359 --> 01:19:11.600 often end up waiting for AI to finish something. 1282 01:19:12.800 --> 01:19:17.960 Yeah, which is you know, doesn't spit things up essentially, Well, yes. 1283 01:19:17.880 --> 01:19:21.319 And no, but but yeah, it's kind of like going 1284 01:19:21.399 --> 01:19:26.199 back to the early two thousands when we had slow compilers. 1285 01:19:27.359 --> 01:19:29.199 Yep, a little bit. 1286 01:19:30.000 --> 01:19:32.439 All right, well, Matteo. If people want to continue to 1287 01:19:32.479 --> 01:19:34.159 follow along with the things you're working on, where do 1288 01:19:34.159 --> 01:19:34.600 they find it. 1289 01:19:35.479 --> 01:19:41.199 They find me on Twitter at mate Hoolina, or they 1290 01:19:41.239 --> 01:19:43.520 can subscribe to my newsletter and. 1291 01:19:44.880 --> 01:19:49.039 At nodland dot dev and yeah, that's kind of. 1292 01:19:48.960 --> 01:19:52.800 It cool, and of course follow at Clubformatic and contact 1293 01:19:52.880 --> 01:19:54.239 us if you need any help. 1294 01:19:54.720 --> 01:19:55.680 Very happy to chat. 1295 01:19:56.159 --> 01:19:57.680 Sounds good, all right. 1296 01:19:57.920 --> 01:20:01.960 Support the people who are making no literally work. 1297 01:20:03.720 --> 01:20:07.399 Yeah all right, yeah, we'll go ahead and wrap up 1298 01:20:07.800 --> 01:20:15.680 until next time. That's out. Mm hmm.