1 00:00:00,120 --> 00:00:02,940 I wouldn't want to program in any other language. 2 00:00:02,940 --> 00:00:07,050 But see, I, I just like the simplicity of the language. 3 00:00:07,080 --> 00:00:09,840 I don't like, uh, the fancy interfaces and I 4 00:00:09,840 --> 00:00:12,900 don't like, you know, things being hidden from me. 5 00:00:12,930 --> 00:00:14,220 So, Pirate's favorite language. 6 00:00:14,220 --> 00:00:17,400 Some people think it might be R but, uh, Pirate's First Love is the c. 7 00:00:20,990 --> 00:00:22,939 The dad jokes, do not get better from here. 8 00:00:27,920 --> 00:00:29,900 Welcome to Screaming in the Cloud. 9 00:00:30,140 --> 00:00:31,400 I'm Corey Quinn. 10 00:00:31,700 --> 00:00:37,910 My guest today is Roi Lipman, who is the CTO at Falco db. 11 00:00:38,330 --> 00:00:39,920 This episode is sponsored in part 12 00:00:39,920 --> 00:00:40,010 by 13 00:00:40,010 --> 00:00:41,810 my day Job Duck Bill. 14 00:00:42,035 --> 00:00:44,614 Do you have a horrifying AWS bill? 15 00:00:44,915 --> 00:00:46,775 That can mean a lot of things. 16 00:00:46,985 --> 00:00:50,074 Predicting what it's going to be, determining what it 17 00:00:50,074 --> 00:00:53,885 should be, negotiating your next long-term contract with 18 00:00:53,885 --> 00:00:57,875 AWS, or just figuring out why it increasingly resembles of. 19 00:00:58,400 --> 00:01:02,030 Phone number, but nobody seems to quite know why that is. 20 00:01:02,330 --> 00:01:05,900 To learn more, visit duck bill hq.com. 21 00:01:06,200 --> 00:01:09,080 Remember, you can't duck the duck bill. 22 00:01:09,110 --> 00:01:13,429 Bill, which my CEO reliably informs me is absolutely 23 00:01:13,429 --> 00:01:14,479 not our slogan. 24 00:01:14,970 --> 00:01:16,500 Chloe, thank you for joining me. 25 00:01:16,590 --> 00:01:18,509 Thank you, co. Happy to be here. 26 00:01:18,810 --> 00:01:20,820 So let's start at the very beginning. 27 00:01:20,850 --> 00:01:22,770 Uh, it seems like you can't swing a dead cat these 28 00:01:22,770 --> 00:01:26,970 days without hitting a different kind of database in 29 00:01:27,149 --> 00:01:31,560 all shapes, sizes, levels of management, et cetera. 30 00:01:32,160 --> 00:01:34,979 What, what type of databases, alco, where does it start? 31 00:01:34,979 --> 00:01:35,610 Where does it stop? 32 00:01:35,670 --> 00:01:39,449 I think, um, over the years there, there was an 33 00:01:39,449 --> 00:01:43,020 explosion of different, uh, databases coming. 34 00:01:43,470 --> 00:01:44,760 Uh, to the market. 35 00:01:44,790 --> 00:01:50,250 Faculty b is a player in the graph database field. 36 00:01:50,880 --> 00:01:53,160 Um, one of many. 37 00:01:53,340 --> 00:01:57,870 Um, you can usually, you know, there are the 38 00:01:57,870 --> 00:02:01,230 relational databases, which everybody probably know. 39 00:02:01,740 --> 00:02:08,729 And then about maybe a decade or two ago, um, we started 40 00:02:08,729 --> 00:02:12,900 to see no SQL databases such as Redis Key Value Store. 41 00:02:13,710 --> 00:02:15,300 Um, of their type. 42 00:02:15,450 --> 00:02:21,930 And within this NoSQL category, you can find document, uh, document 43 00:02:21,930 --> 00:02:27,630 databases such as, uh, Mongo, you see time series databases. 44 00:02:27,720 --> 00:02:30,990 Today you would also find vector databases. 45 00:02:31,470 --> 00:02:37,350 And there's this niche, um, which is the, you know, 46 00:02:37,350 --> 00:02:41,520 graph database field, uh, which we've been in. 47 00:02:42,390 --> 00:02:44,400 For the past 10 years now, 48 00:02:44,490 --> 00:02:48,810 I'm reminded of Forrest Brazil's comic years ago before he joined Google. 49 00:02:49,020 --> 00:02:51,825 Uh, sort of disambiguating all of the different. 50 00:02:52,545 --> 00:02:54,885 AWS managed database services. 51 00:02:55,005 --> 00:02:58,095 'cause it really seemed for a little while there, like the DBA job of the 52 00:02:58,095 --> 00:03:01,725 future was gonna be figuring out which of their 40 databases was the right 53 00:03:01,725 --> 00:03:06,465 option for any given workload And for, uh, the Neptune option that they offered. 54 00:03:06,585 --> 00:03:10,785 The, uh, the flow chart decision was, do you need, uh, graph database? 55 00:03:10,785 --> 00:03:12,885 And the only option was No, you don't. 56 00:03:12,885 --> 00:03:15,945 And it led right back to the previous question, which I thought was. 57 00:03:16,395 --> 00:03:17,415 Pretty amusing. 58 00:03:17,745 --> 00:03:20,715 Uh, it seems these days that graph databases 59 00:03:20,715 --> 00:03:22,815 have found a niche, but, but it is a niche. 60 00:03:22,815 --> 00:03:25,665 It's, it's a, it's, I've not built anything on top of one yet. 61 00:03:25,935 --> 00:03:30,585 And I talking to other folks, I don't think I'm very alone in that. 62 00:03:30,644 --> 00:03:31,755 Am I, am I the freak? 63 00:03:32,115 --> 00:03:33,465 No, I don't think so. 64 00:03:33,465 --> 00:03:38,355 I think that, uh, it's very natural for, to, for people to think. 65 00:03:38,805 --> 00:03:39,795 In tables. 66 00:03:39,915 --> 00:03:43,785 Well, it comes very natural to say, oh, my data fits into 67 00:03:43,785 --> 00:03:48,644 a table, so I'm gonna go with relational, uh, database. 68 00:03:49,065 --> 00:03:56,954 Um, those who are aware of this entity called Graph, uh, these are 69 00:03:56,954 --> 00:04:02,834 either, um, people who have encountered this probably during their time. 70 00:04:03,150 --> 00:04:04,350 At the university 71 00:04:04,680 --> 00:04:05,760 or working at Facebook 72 00:04:06,090 --> 00:04:13,380 or, yeah, I mean, if they think about, you know, how uh, the data in Facebook 73 00:04:13,380 --> 00:04:19,680 is modeled then, or in other, in any other social network, you would see this 74 00:04:19,680 --> 00:04:27,450 shape of a network or more generally We like to call it a graph forms, so, um. 75 00:04:28,020 --> 00:04:31,200 Apparently, uh, or as it, as it turns out, there is 76 00:04:31,200 --> 00:04:35,580 this very famous saying, uh, everything is a graph. 77 00:04:35,940 --> 00:04:46,260 So even if you are aware of it or not, um, it's very likely that a lot of 78 00:04:46,260 --> 00:04:54,480 domains are behind the scene or can be relatively easy described as a graph. 79 00:04:54,600 --> 00:04:57,060 Just, just throwing out a few examples. 80 00:04:57,615 --> 00:05:05,325 A social network obviously is a graph, roads and intersections, maps. 81 00:05:05,835 --> 00:05:07,185 These are graphs. 82 00:05:07,485 --> 00:05:16,335 Any type of, um, uh, network where data flows is, is a graph. 83 00:05:17,295 --> 00:05:21,645 Um, so graph do pop up, uh, in a lot of use cases. 84 00:05:21,645 --> 00:05:23,985 It's, it's just, uh, the. 85 00:05:24,765 --> 00:05:28,425 You know, the way in which we're viewing the problem and 86 00:05:28,905 --> 00:05:33,525 realizing that, oh, uh, I'm dealing with a graph here. 87 00:05:34,185 --> 00:05:40,844 So maybe instead of trying to false a graph structure into a different, 88 00:05:40,965 --> 00:05:47,205 uh, data model, such as the document store or a relational database 89 00:05:47,205 --> 00:05:52,724 with tables, maybe I should use the right data model for my domain. 90 00:05:53,295 --> 00:05:54,465 Or for my problem. 91 00:05:54,615 --> 00:06:00,165 And just to, to wrap this up, think about a navigation system. 92 00:06:00,195 --> 00:06:09,615 The one that you might have in your car whenever somebody changes, uh, the map. 93 00:06:09,675 --> 00:06:16,275 Let's say a new road was added, or uh, some road was removed 94 00:06:16,275 --> 00:06:19,425 because there's some construction and then the road is closed. 95 00:06:20,700 --> 00:06:25,380 Every update to that map need to go through a validation that 96 00:06:25,380 --> 00:06:30,419 says, uh, for example, in a given, if in a given state or 97 00:06:30,419 --> 00:06:35,849 in a given country, you should be able to drive from every 98 00:06:36,270 --> 00:06:42,419 location to any other location on that map or within that state. 99 00:06:42,510 --> 00:06:44,250 That is definition of how roads should work. 100 00:06:44,309 --> 00:06:44,580 Yeah. 101 00:06:44,580 --> 00:06:45,599 And it's pretty obvious. 102 00:06:45,599 --> 00:06:51,599 I mean, you should get from point A to point B. Try to solve this 103 00:06:51,630 --> 00:06:58,799 problem or make this verification when modeling this data as a table or 104 00:06:58,859 --> 00:07:03,929 as a set of tables or a set of documents, it's practically impossible. 105 00:07:04,140 --> 00:07:05,549 You need, you need a graph. 106 00:07:05,549 --> 00:07:08,940 It's a, it's a graph problem to begin with. 107 00:07:09,630 --> 00:07:17,190 And so, yes, I think that graphs are still a niche, but they do, uh, with. 108 00:07:17,760 --> 00:07:21,450 The explosion of ai, we do see them coming back. 109 00:07:22,185 --> 00:07:25,635 Real strong with this idea of knowledge graphs. 110 00:07:25,635 --> 00:07:30,885 For a while my default go-to was DynamoDB, and that was great for a 111 00:07:30,885 --> 00:07:33,645 variety of reasons for small throwaway projects, but something you 112 00:07:33,645 --> 00:07:36,825 learn pretty quickly when you start reaching for, uh, the every, uh. 113 00:07:37,290 --> 00:07:39,060 Every, every, you know, the tool you have is a hammer. 114 00:07:39,060 --> 00:07:40,230 Every problem looks like your thumb. 115 00:07:40,590 --> 00:07:43,560 Uh, when you're using Dynamo, it works really well. 116 00:07:43,560 --> 00:07:46,080 If and only if you know exactly what your access 117 00:07:46,080 --> 00:07:48,180 patterns are going to look like in advance. 118 00:07:48,450 --> 00:07:52,170 As soon as you start changing that due to evolving requirements, 119 00:07:52,290 --> 00:07:55,320 suddenly you're in a world of pain because you get to reimagine your 120 00:07:55,320 --> 00:07:58,560 entire table model, redo a bunch of stuff, possibly do a migration. 121 00:07:58,830 --> 00:08:02,910 So I have found that reaching by default for something like Postgre 122 00:08:02,910 --> 00:08:06,780 squeal tends to be the right answer for a lot of these things. 123 00:08:07,120 --> 00:08:10,570 But also increasingly I find that I don't care what 124 00:08:10,570 --> 00:08:13,330 database it uses 'cause I'm not the one building the code. 125 00:08:13,330 --> 00:08:18,370 I am hurling it over the wall for an AI tool to wind up building out for me. 126 00:08:18,820 --> 00:08:23,440 Uh, how are you seeing, I guess, the rise of graph in that universe? 127 00:08:23,770 --> 00:08:24,775 In the universe of. 128 00:08:25,315 --> 00:08:27,715 People building applications with ai, 129 00:08:27,715 --> 00:08:29,695 throwing it over the wall to AI so that it 130 00:08:29,695 --> 00:08:31,765 just decides whatever it's going to build. 131 00:08:31,765 --> 00:08:34,195 I, it has preferences and in some cases I've switched 132 00:08:34,195 --> 00:08:37,495 providers just because I got tired of fighting against it. 133 00:08:37,495 --> 00:08:40,315 Like, no use Amplify, don't want, I want to use Versal Fine. 134 00:08:40,315 --> 00:08:40,885 Use Versal. 135 00:08:40,885 --> 00:08:42,294 Just stop bothering me with it. 136 00:08:42,534 --> 00:08:51,055 I personally, personally, try to avoid handing off my work, uh, to ai. 137 00:08:51,175 --> 00:08:53,995 So I use it mostly for. 138 00:08:54,540 --> 00:08:55,709 Doing code review. 139 00:08:56,459 --> 00:09:02,219 So I, I don't have enough experience to to, to answer that question. 140 00:09:02,219 --> 00:09:05,339 I, I'm not in the position where I constantly 141 00:09:05,880 --> 00:09:08,910 hand up handing off, handing off work to an ai. 142 00:09:09,360 --> 00:09:09,420 So 143 00:09:10,500 --> 00:09:10,800 Right. 144 00:09:10,800 --> 00:09:13,770 In fairness, that's, that is where most developers are. 145 00:09:13,770 --> 00:09:16,620 I have the advantage of being a different kind of developer where 146 00:09:16,620 --> 00:09:19,740 my only two languages I know are brute force and enthusiasm. 147 00:09:20,010 --> 00:09:24,180 When I want a line of business app to glue two things together, well, that 148 00:09:24,180 --> 00:09:27,125 would've been a multi-day undertaking previously, and now it's, eh, just. 149 00:09:27,400 --> 00:09:29,590 There'll be a clicky thing so I can stop copying and 150 00:09:29,590 --> 00:09:31,870 pasting between two things like I have for four years. 151 00:09:32,170 --> 00:09:36,130 It's, it's basically the glue applications between things. 152 00:09:36,490 --> 00:09:41,890 And historically I found that I cared a lot more about things like 153 00:09:41,890 --> 00:09:45,310 data structures, the underlying code, how this stuff all ties together. 154 00:09:45,579 --> 00:09:47,680 Increasingly, I find as long as I have a solid 155 00:09:47,680 --> 00:09:50,140 test harness, I can, I don't have to care as much. 156 00:09:50,260 --> 00:09:53,439 Now, will this have problems if I start scaling any of these things? 157 00:09:53,620 --> 00:09:54,940 Oh, absolutely. 158 00:09:55,180 --> 00:09:58,360 But that's not a, that's not a risk for the way that 159 00:09:58,360 --> 00:10:00,880 I am building a lot of these things to begin with. 160 00:10:01,150 --> 00:10:04,180 It's more or less building everything with the idea from the outset 161 00:10:04,180 --> 00:10:06,460 that it's probably not gonna work and I'll throw it away and replace 162 00:10:06,460 --> 00:10:09,670 it in two months with something else, which is frankly, a luxury. 163 00:10:10,090 --> 00:10:12,130 But I do find that it gets me experimenting with 164 00:10:12,130 --> 00:10:14,740 different technologies I otherwise would not have. 165 00:10:15,315 --> 00:10:20,320 Had the ramp up time or focus to dive, really dive into, have 166 00:10:20,320 --> 00:10:23,205 you seen, have you seen folks that is coming to Falco from, 167 00:10:23,325 --> 00:10:26,415 well, AI suggested it, so now I'm curious, or is that not 168 00:10:26,415 --> 00:10:29,265 really the, the path that leads folks to a graph database? 169 00:10:29,385 --> 00:10:32,805 The way that our prospects of customers are ending 170 00:10:32,805 --> 00:10:36,975 up using Falco D Beta are usually two main paths. 171 00:10:37,275 --> 00:10:39,975 It's either they realize that they need a 172 00:10:39,975 --> 00:10:43,365 graph and they, they do a quick, uh, search. 173 00:10:43,709 --> 00:10:47,339 And they, uh, come up with a, a number of contends 174 00:10:48,300 --> 00:10:51,630 and, uh, hopefully they would pick a faculty. 175 00:10:52,230 --> 00:10:55,020 So those are sometimes they, they are, 176 00:10:55,829 --> 00:10:59,880 uh, coming fresh to this type of database. 177 00:10:59,880 --> 00:11:04,469 They did not have experience before, but they understand that the problem 178 00:11:04,469 --> 00:11:09,660 that they're trying to solve is best or is suited for a graph database. 179 00:11:10,050 --> 00:11:11,219 That's one path. 180 00:11:12,495 --> 00:11:18,135 The other one is, uh, prospects who had some experience with 181 00:11:18,135 --> 00:11:22,605 the grab database, maybe a short while or even for a few years. 182 00:11:23,295 --> 00:11:28,755 Um, and for any, you know, there might be multiple reasons 183 00:11:28,965 --> 00:11:32,145 that they're not happy with their current provider. 184 00:11:32,805 --> 00:11:38,505 Um, and so they, they want to try something else, usually. 185 00:11:39,135 --> 00:11:45,315 Uh, those people who are turning to fle DB are turning in, uh, 186 00:11:45,585 --> 00:11:52,395 because they have concerns regard to either scaling or performance. 187 00:11:52,395 --> 00:11:55,005 So this is where fle DB shines. 188 00:11:55,425 --> 00:12:01,185 Um, when you try to compare it to the other vendors, um, we have a unique 189 00:12:01,185 --> 00:12:06,465 approach to how we go about representing graphs and traversing them. 190 00:12:06,915 --> 00:12:12,555 But we also make it extremely easy for you to scale out, uh, in case 191 00:12:12,555 --> 00:12:16,560 you have multiple tenants or multiple graphs that you want to host. 192 00:12:17,430 --> 00:12:20,010 On your, uh, Falco DB database. 193 00:12:20,069 --> 00:12:23,670 Something that I, I noticed about Falco that ties into, I guess, a 194 00:12:23,670 --> 00:12:27,540 preexisting bias I've had because for a number of years it seems that 195 00:12:27,540 --> 00:12:31,920 folks have been pushing really hard to make Vector Search a its own 196 00:12:31,920 --> 00:12:36,360 database type, but it, it's become pretty clear across the board, 197 00:12:36,360 --> 00:12:38,729 across these are the database types that I spend more time with. 198 00:12:38,969 --> 00:12:41,699 That vector search is a feature, it is 199 00:12:41,699 --> 00:12:45,000 not an engine itself, and it appears that. 200 00:12:45,740 --> 00:12:49,970 Falcor has, Falco DB has taken this in a similar direction where you, you 201 00:12:49,970 --> 00:12:53,360 obviously have done a fair bit of enrichment with, uh, your own graph rag 202 00:12:53,360 --> 00:12:58,310 stuff, but vector search is a, is a already a feature that Falco DB does. 203 00:12:58,700 --> 00:12:59,750 Where do you land on that 204 00:13:00,470 --> 00:13:06,170 vector search or semantic search is, um, really. 205 00:13:06,570 --> 00:13:11,880 Important type of search that you, you can perform at the very 206 00:13:11,880 --> 00:13:15,660 beginning, or at least this is how I remember things a few years 207 00:13:15,660 --> 00:13:23,010 back with the, with the, with the ability to compute, embeddings 208 00:13:23,730 --> 00:13:30,570 and then search, given some data point, which you also turn into 209 00:13:30,570 --> 00:13:35,610 a, a, a vector, and then you search through the, the vector space. 210 00:13:36,194 --> 00:13:43,244 Uh, the first product that came out were actually vector databases. 211 00:13:43,244 --> 00:13:47,175 Companies such as Pine Cone, wate, I think 212 00:13:47,444 --> 00:13:50,204 Quadrant, and there, there are plenty. 213 00:13:51,074 --> 00:13:57,135 Um, and eventually the other players, relational databases 214 00:13:57,135 --> 00:14:01,214 of databases, any type of database looked at it as if. 215 00:14:01,949 --> 00:14:03,630 I don't think I want this feature. 216 00:14:03,630 --> 00:14:10,140 I want the ability to perform semantic search over my own, you know, data. 217 00:14:10,140 --> 00:14:14,520 I don't need a, a, a new database to add to the stack and manage. 218 00:14:15,720 --> 00:14:18,449 I look at it as a, as a search capability. 219 00:14:18,480 --> 00:14:23,655 And we already have indices doing full text, doing exact match. 220 00:14:24,875 --> 00:14:27,689 Why not doing also, uh, geo. 221 00:14:28,485 --> 00:14:33,465 Why not ex extend the index capability with, with Symantec? 222 00:14:34,215 --> 00:14:36,525 And this is what the entire industry had done. 223 00:14:36,525 --> 00:14:42,915 You would see these days, um, throughout the spectrum, almost 224 00:14:42,915 --> 00:14:48,345 any, any database would now support, uh, a vector capability. 225 00:14:48,525 --> 00:14:54,915 Now, I'm not saying that there's no no room for vector databases out there. 226 00:14:54,944 --> 00:14:57,525 I mean, if you are hosting. 227 00:14:58,230 --> 00:15:07,170 Uh, hundreds of, of millions or, uh, or even more vectors and that's, 228 00:15:07,170 --> 00:15:12,750 you know, the only type of information that you are storing for your 229 00:15:12,750 --> 00:15:18,120 application, then yeah, the vector database, you know, might make sense. 230 00:15:18,300 --> 00:15:19,590 Um, but. 231 00:15:20,775 --> 00:15:27,075 If your numbers are relatively, you know, reasonable, and you need those 232 00:15:27,075 --> 00:15:34,245 vectors just to do semantic search over the broader, uh, data set that you have. 233 00:15:34,485 --> 00:15:38,085 Uh, for example, let's say you have a, a database 234 00:15:38,085 --> 00:15:41,300 containing users and every user has a profile. 235 00:15:41,954 --> 00:15:46,005 Image and then you are getting a new image and you want 236 00:15:46,005 --> 00:15:50,895 to see, you know, which the top five users that are, 237 00:15:51,194 --> 00:15:54,735 you know, they look like the, the given query image. 238 00:15:55,035 --> 00:16:00,285 Then okay, use, use a vector index capability. 239 00:16:00,975 --> 00:16:02,055 So that's what we've done. 240 00:16:02,055 --> 00:16:04,245 This is what the entire industry had done. 241 00:16:04,574 --> 00:16:08,355 That, that I think, put a question mark on 242 00:16:08,655 --> 00:16:11,564 whether or not do we really need vector databases. 243 00:16:12,285 --> 00:16:17,415 I think that for the extreme cases, yes, but for the majority you would 244 00:16:17,415 --> 00:16:22,695 see people are using Postgres and they're doing TAL cells within Postgres. 245 00:16:23,025 --> 00:16:26,175 Others are doing it within file cord tb because you know, they 246 00:16:26,175 --> 00:16:29,295 have a graph and then they, they want this semantic capability. 247 00:16:29,579 --> 00:16:32,369 Which tracks, I mean, I'm not here to crop on other people's use cases. 248 00:16:32,400 --> 00:16:36,150 It's, it's one of those areas where people get very sensitive about it. 249 00:16:36,150 --> 00:16:37,050 It's like they're children. 250 00:16:37,050 --> 00:16:38,640 It's like, oh, so which is your favorite child? 251 00:16:38,640 --> 00:16:42,030 I can't stand any of them if I'm being direct now, it's, 252 00:16:42,060 --> 00:16:47,219 it becomes a almost a religious debate at some point. 253 00:16:47,489 --> 00:16:47,579 Mm-hmm. 254 00:16:48,479 --> 00:16:50,280 Where did Falco DB come from? 255 00:16:50,339 --> 00:16:52,709 What is the, what is the origin story of it? 256 00:16:52,739 --> 00:16:55,020 Because frankly, I, one day someone decided 257 00:16:55,020 --> 00:16:56,910 to sit down, I'm gonna write a graph database. 258 00:16:56,939 --> 00:17:00,390 Sounds like a supervi origin story more than anything else. 259 00:17:00,540 --> 00:17:03,060 Well, that's, that's what happened 10 years ago. 260 00:17:03,329 --> 00:17:03,959 Uh, with me. 261 00:17:05,399 --> 00:17:12,179 I was, um, I took a course, uh, at the university that had dealt 262 00:17:12,179 --> 00:17:16,560 with graph, and I just fell in love with this idea of graphs. 263 00:17:17,099 --> 00:17:22,649 Um, and, uh, I thought that back then I thought that it would be 264 00:17:22,649 --> 00:17:29,190 interesting to extend Redis capability with, uh, a graph data type. 265 00:17:29,250 --> 00:17:36,060 Back then Redis just came out with its ability to extend, uh, the database 266 00:17:36,060 --> 00:17:41,580 with modules, read this modules, and I thought it would be a nice idea 267 00:17:41,580 --> 00:17:46,169 to have a graph data type and some graph capabilities within reds. 268 00:17:46,935 --> 00:17:50,595 So, um, I worked on that project for a while. 269 00:17:50,595 --> 00:17:53,715 I think it was about, uh, six months. 270 00:17:53,835 --> 00:17:58,034 And then I met with the, the Redis folks and I, you know, showed them 271 00:17:58,034 --> 00:18:02,445 what I've been up to and they said, oh, that's really interesting. 272 00:18:02,445 --> 00:18:04,905 You should come and join Redis. 273 00:18:05,264 --> 00:18:11,100 Um, so I did, and within Redis I've built Redis graph. 274 00:18:11,445 --> 00:18:13,780 So we turned this six month. 275 00:18:14,745 --> 00:18:19,635 Uh, worth of work at home into an actual product. 276 00:18:20,235 --> 00:18:24,975 Um, I've done that at Redis for about six to seven years, 277 00:18:25,635 --> 00:18:31,304 building and developing Redis graph, um, which eventually 278 00:18:31,544 --> 00:18:36,105 Redis decided to abandon or, you know, shut down the project. 279 00:18:36,824 --> 00:18:40,034 And, uh, myself and two other colleagues of mine 280 00:18:40,034 --> 00:18:42,345 thought that it would be a shame to see this project. 281 00:18:43,065 --> 00:18:44,385 Go down the drain. 282 00:18:44,895 --> 00:18:49,455 And so we decided to spin out of Redis and turn 283 00:18:49,725 --> 00:18:55,095 Redis graph into what now is known as Fal Cord db. 284 00:18:55,095 --> 00:19:00,315 So we've basically just continue on from where we've left off at Redis. 285 00:19:00,855 --> 00:19:05,145 Um, and we've been doing that for over two years now. 286 00:19:05,385 --> 00:19:08,415 So around 10 years in total. 287 00:19:08,504 --> 00:19:11,450 I'm involved in building and, you know. 288 00:19:12,225 --> 00:19:17,145 Getting a graph database out there, which is competitive. 289 00:19:17,505 --> 00:19:21,795 Um, so yeah, that's, that's the background story behind File Core. 290 00:19:23,115 --> 00:19:26,655 This episode is sponsored by my own company, duck Bill. 291 00:19:26,925 --> 00:19:29,895 Having trouble with your AWS bill, perhaps 292 00:19:29,895 --> 00:19:32,565 it's time to renegotiate a contract with them. 293 00:19:32,925 --> 00:19:35,055 Maybe you're just wondering how to predict 294 00:19:35,055 --> 00:19:38,325 what's going on in the wide world of AWS. 295 00:19:38,355 --> 00:19:40,995 Well, that's where Duck Bill comes to help. 296 00:19:41,205 --> 00:19:43,965 Remember, you can't duck the duck bill. 297 00:19:43,965 --> 00:19:45,765 Bill, which I am reliably 298 00:19:45,765 --> 00:19:50,145 informed by my business partner is absolutely not our motto. 299 00:19:50,405 --> 00:19:53,675 To learn more, visit bill hq com. 300 00:19:54,425 --> 00:19:57,995 How do you find that it has been to run effectively? 301 00:19:57,995 --> 00:20:01,455 What is a. I, I, I have to be careful using terms like open 302 00:20:01,455 --> 00:20:04,034 source, people get salty about this, but you are source available. 303 00:20:04,245 --> 00:20:09,225 Uh, do you find that it is tricky to run a viable business where 304 00:20:09,225 --> 00:20:12,465 there's also the story of, or you can just use the code and all 305 00:20:12,465 --> 00:20:15,824 the stuff that we build for free and make available to everyone? 306 00:20:16,040 --> 00:20:18,465 I, I, I think I understand that it's somewhat of a leading 307 00:20:18,465 --> 00:20:21,135 question and a naive one, but I am curious to get your take on it. 308 00:20:21,344 --> 00:20:23,175 Yeah, I mean, it's tricky. 309 00:20:23,235 --> 00:20:24,735 It's a question that pops up. 310 00:20:25,020 --> 00:20:29,070 Uh, from time to time, I think that the, the benefits of having your 311 00:20:29,070 --> 00:20:34,199 product as an open source are greater than having it as a closed source. 312 00:20:34,590 --> 00:20:41,129 Um, what happens is that your adoption rate is much, is much greater when 313 00:20:41,129 --> 00:20:48,090 you have your product, um, available for free on the other, on the other end. 314 00:20:48,090 --> 00:20:50,100 Yes, it's a bit frustrating to see all of 315 00:20:50,100 --> 00:20:51,915 those people who are using the product and. 316 00:20:53,010 --> 00:20:55,620 Are not chipping, chipping in. 317 00:20:56,129 --> 00:20:59,940 Uh, but I mean, at the end of the day, for me, it's a, 318 00:20:59,940 --> 00:21:05,250 it's a huge, um, it's, it's just a pleasure to see, uh, 319 00:21:05,250 --> 00:21:09,810 the work that I've done being used all over the place. 320 00:21:09,870 --> 00:21:12,629 And it, it comes as a surprise. 321 00:21:12,629 --> 00:21:15,060 You know, we, we are not aware of all of the use 322 00:21:15,060 --> 00:21:17,370 cases and all of the people who are using the product. 323 00:21:17,370 --> 00:21:19,110 And when somebody reaches out. 324 00:21:20,129 --> 00:21:23,100 Uh, and we, you know, have a conversation with him and you see the 325 00:21:23,100 --> 00:21:28,500 product being used for a variety of, uh, workloads and use cases. 326 00:21:28,919 --> 00:21:32,159 It, you know, you feel, you feel great about that. 327 00:21:32,669 --> 00:21:37,139 In addition, the fact that it, it is source available, people 328 00:21:37,139 --> 00:21:41,790 can submit issues, people can have a conversation about it. 329 00:21:42,449 --> 00:21:45,300 Um, there is, there is benefit to it. 330 00:21:45,389 --> 00:21:47,850 So if you look at our Gita page, you'll see. 331 00:21:48,780 --> 00:21:50,760 Um, plenty of issues. 332 00:21:50,939 --> 00:21:55,350 Um, and I think it's, it's great to have GitHub issues because 333 00:21:55,350 --> 00:21:57,989 it shows that people are, are using the product and they're, 334 00:21:58,110 --> 00:22:02,550 and, and they care enough to go and, and open an issue. 335 00:22:03,149 --> 00:22:05,879 Um, and so we, we learn from that. 336 00:22:05,879 --> 00:22:11,850 I mean, it's kind of like a free QA to the product we might be able 337 00:22:11,850 --> 00:22:16,949 to produce or to generate more revenue if we were closed source. 338 00:22:17,325 --> 00:22:20,175 But I think that the, the benefits of having something open 339 00:22:20,175 --> 00:22:23,790 and making some contribution to the open source community. 340 00:22:24,840 --> 00:22:29,370 Um, does pay off, uh, at the, at the long climb, 341 00:22:29,489 --> 00:22:30,120 apparently. 342 00:22:30,179 --> 00:22:35,040 Uh, back in 2024, you folks did a, uh, an 343 00:22:35,040 --> 00:22:38,399 announcement about rewriting Alcore DB and rust. 344 00:22:38,429 --> 00:22:38,520 Mm-hmm. 345 00:22:38,909 --> 00:22:39,959 Uh, that often. 346 00:22:39,959 --> 00:22:43,560 I, my snide comment has always been the primary IDE of a rust 347 00:22:43,560 --> 00:22:46,229 developer is PowerPoint because it's impossible to do anything 348 00:22:46,229 --> 00:22:48,540 in rust without doing a whole dog and pony show around it. 349 00:22:48,780 --> 00:22:49,770 Uh, has that been completed? 350 00:22:49,770 --> 00:22:50,189 How'd it go? 351 00:22:50,310 --> 00:22:52,290 No, it's still a walk in progress. 352 00:22:52,939 --> 00:22:58,399 Um, my other, uh, co-founder Avi, is in 353 00:22:58,399 --> 00:23:02,149 charge of doing the, the vast right at first. 354 00:23:02,149 --> 00:23:05,689 We've tried to do that in a gradual form, so 355 00:23:05,689 --> 00:23:09,949 we have this relatively large C code base. 356 00:23:09,980 --> 00:23:10,790 I'm a big. 357 00:23:11,324 --> 00:23:12,524 C fan. 358 00:23:12,584 --> 00:23:18,824 I, I am a, I wouldn't want to program in any other language, but C Oh, 359 00:23:19,245 --> 00:23:21,405 that, that'll get you hate mail just for that statement alone. 360 00:23:22,814 --> 00:23:25,185 Uh, maybe I, I mean, I'm 42 years old. 361 00:23:25,185 --> 00:23:29,895 Um, so maybe that's the, you know, the, the age. 362 00:23:30,095 --> 00:23:34,475 Yeah, we are a gentleman of a certain age and yeah, at some point learning the 363 00:23:34,475 --> 00:23:39,095 latest new, trendy things, it just feels like it's more trouble than it's worth. 364 00:23:39,095 --> 00:23:40,325 If I'm being direct, 365 00:23:40,475 --> 00:23:43,595 I, I just like the simplicity of the language. 366 00:23:43,625 --> 00:23:46,415 I don't like, uh, the fancy interfaces and I 367 00:23:46,415 --> 00:23:49,355 don't like, you know, things being hidden from me. 368 00:23:49,685 --> 00:23:53,254 The type of programming that it is needed for, to develop a high 369 00:23:53,254 --> 00:23:58,085 performance curve database, you need to be aware or, and in control of. 370 00:23:59,130 --> 00:24:02,910 Practically anything that happens in your system and any language that 371 00:24:02,970 --> 00:24:08,520 hides stuff from you, uh, and the language that would not give you 372 00:24:08,520 --> 00:24:15,270 absolute control over the instructions and the memory layout, you would 373 00:24:15,270 --> 00:24:19,590 simply, you know, would not be able to fully utilize the hardware. 374 00:24:19,680 --> 00:24:22,590 Just circling back to the Russ question, so 375 00:24:22,590 --> 00:24:25,530 we, we, the first, we try to do it gradually. 376 00:24:26,175 --> 00:24:30,630 Taking apart, uh, components that are ancy and reit them in 377 00:24:31,065 --> 00:24:35,655 and integrating them back, that didn't work too well for us. 378 00:24:35,955 --> 00:24:39,735 And so, uh, after a year of trying this out, 379 00:24:39,795 --> 00:24:42,735 we've decided that this approach would not work. 380 00:24:42,825 --> 00:24:47,445 And so we're gonna write everything in one go the entire thing from Sketch. 381 00:24:47,685 --> 00:24:52,035 So that, that's gonna take a while in my mind. 382 00:24:53,025 --> 00:24:56,535 Um, they're, they're due, you know, they did make a 383 00:24:56,535 --> 00:25:00,795 good progress, uh, but it's still not there just yet. 384 00:25:01,935 --> 00:25:04,905 Um, I'm more focused on the C, 385 00:25:05,205 --> 00:25:06,465 so Pirate's favorite language. 386 00:25:06,465 --> 00:25:09,044 Some people think it might be R, but the Pirate's first. 387 00:25:09,044 --> 00:25:09,735 Love is the C 388 00:25:13,395 --> 00:25:13,635 Dad. 389 00:25:13,635 --> 00:25:15,135 Jokes do not get better from here. 390 00:25:16,635 --> 00:25:18,014 I store them in a database. 391 00:25:18,290 --> 00:25:20,480 There's a separate, there's a whole spiel we 392 00:25:20,480 --> 00:25:23,149 can go into and we'll avoid for brevity's sake. 393 00:25:23,210 --> 00:25:27,050 What I guess is the, the challenge I've always had, and I suspect I'm not 394 00:25:27,050 --> 00:25:31,490 alone in this, is I keep seeing graph databases and I see things that are 395 00:25:31,490 --> 00:25:36,810 interesting, but I, I'm missing a toy problem that feels like a good fit for it. 396 00:25:36,824 --> 00:25:37,044 Um. 397 00:25:37,980 --> 00:25:43,320 Take a relational database and great, I'm going to go and build even 398 00:25:43,320 --> 00:25:46,500 something dumb like a to-do app, which is still overkill for some of that. 399 00:25:46,740 --> 00:25:50,370 Or I'm going to build something that will track books in a library, 400 00:25:50,400 --> 00:25:53,520 uh, with, and you get multi table stories with authors and books. 401 00:25:53,520 --> 00:25:57,510 And the rest I, what is, I guess, the toy problem, uh, minimum 402 00:25:57,510 --> 00:26:01,170 viable footprint that isn't insane for a graph database. 403 00:26:01,320 --> 00:26:03,750 So I have, uh, maybe two examples. 404 00:26:03,750 --> 00:26:06,270 We've recently done a webinar about. 405 00:26:06,855 --> 00:26:12,315 Our UDF capabilities, user, user defined functions, uh, within the database. 406 00:26:12,375 --> 00:26:18,885 And, uh, the, um, project which showcased this was farm to table. 407 00:26:19,245 --> 00:26:26,085 So the idea was in this graph you have nodes, entities representing forms. 408 00:26:26,475 --> 00:26:28,650 Those are connected to delivery hubs. 409 00:26:29,595 --> 00:26:32,415 Which might connect to other delivery hubs, but 410 00:26:32,415 --> 00:26:36,195 eventually they would reach a store like Trader Joe's. 411 00:26:36,915 --> 00:26:42,495 And basically what you want to do is to find a path that would get the 412 00:26:42,495 --> 00:26:49,425 produce fruits and vegetables and keep them as fresh as possible to the store. 413 00:26:49,515 --> 00:26:51,945 So in other words, it's an implementation of the traveling salesman problem. 414 00:26:52,035 --> 00:26:55,665 Yes, you can You, I mean, you can think about it that way. 415 00:26:55,725 --> 00:26:57,495 It's a path finding problem. 416 00:26:58,020 --> 00:27:02,129 Where you want to optimize, uh, some function in, 417 00:27:02,220 --> 00:27:05,700 in our case, it was the, the freshness of the food. 418 00:27:06,330 --> 00:27:11,160 Uh, but you can also look at it and say, okay, I want to, uh, 419 00:27:11,190 --> 00:27:16,320 find the path which has the minimum cost of transportation. 420 00:27:16,649 --> 00:27:20,610 So that's maybe one example where, you know, 421 00:27:20,610 --> 00:27:24,750 we got databases, uh, are, are a great fit. 422 00:27:24,750 --> 00:27:26,385 Another one, which, you know, people. 423 00:27:26,745 --> 00:27:29,564 It can relate to is access management. 424 00:27:30,014 --> 00:27:34,004 So think about a big organization with lots of users and those users 425 00:27:34,004 --> 00:27:38,054 are, uh, grouped into groups and there might be a group hierarchy. 426 00:27:38,060 --> 00:27:43,125 And then there are resources such as servers and files and folders. 427 00:27:43,635 --> 00:27:46,574 And every time a user tried to access one of those 428 00:27:46,574 --> 00:27:50,205 resources, you need to make sure that he has the right axis. 429 00:27:50,264 --> 00:27:54,165 And so this once again becomes a, a path finding problem. 430 00:27:54,165 --> 00:27:56,189 If you think of it, you need to see if there is a path. 431 00:27:57,345 --> 00:28:01,275 Connecting that user through a set of different 432 00:28:01,515 --> 00:28:05,445 groups to the end, uh, to the end resource. 433 00:28:05,625 --> 00:28:07,260 So we actually have a number of. 434 00:28:08,235 --> 00:28:10,545 Customers who are using Falco tb Exactly. 435 00:28:10,545 --> 00:28:11,295 For this, 436 00:28:11,385 --> 00:28:14,355 my last question before we wind up calling this an 437 00:28:14,355 --> 00:28:17,505 episode is open source, source available, et cetera. 438 00:28:17,505 --> 00:28:17,865 Great. 439 00:28:17,925 --> 00:28:18,345 Awesome. 440 00:28:18,405 --> 00:28:22,335 Your code is on GitHub and accepts contributions from the larger community. 441 00:28:23,085 --> 00:28:26,085 I have heard a lot of open source maintainers. 442 00:28:26,145 --> 00:28:26,505 Uh. 443 00:28:26,679 --> 00:28:28,600 I have some minor stuff up, but you know, we're talking 444 00:28:28,600 --> 00:28:30,250 to actual people who are serious about this stuff. 445 00:28:30,490 --> 00:28:36,970 Uh, they are inundated with a barrage of AI generated pull requests. 446 00:28:37,240 --> 00:28:42,580 The quality of which is, uh, will be polite and say varies widely. 447 00:28:43,149 --> 00:28:44,860 What has your experience been with that 448 00:28:45,100 --> 00:28:45,700 faculty? 449 00:28:45,700 --> 00:28:50,860 B is is not an easy product to contribute code to. 450 00:28:51,645 --> 00:28:56,145 So if we're getting some contributions, it's, it's 451 00:28:56,145 --> 00:28:59,145 very rare and we're very happy to receive them. 452 00:28:59,895 --> 00:29:05,055 That said, a few, a few years back before, uh, you 453 00:29:05,055 --> 00:29:08,925 know, all of those new AI tools, whenever a PR showed. 454 00:29:09,255 --> 00:29:13,545 Uh, it was obvious that the, there was a person behind that PI 455 00:29:14,175 --> 00:29:19,395 who took the time and wrote the, the piece of code that he wanted, 456 00:29:19,545 --> 00:29:19,785 right? 457 00:29:19,785 --> 00:29:21,015 Someone cared. 458 00:29:21,075 --> 00:29:22,995 Uh, my fir, one of my early exposures to the open 459 00:29:22,995 --> 00:29:25,005 source world in that sense was with SaltStack. 460 00:29:25,305 --> 00:29:28,545 Tom Hatch was a phenomenal founder of that 461 00:29:28,545 --> 00:29:30,660 project and company 'cause he would accept. 462 00:29:31,264 --> 00:29:36,635 My shitty poll request, and then he would thank me for it and then immediately 463 00:29:36,784 --> 00:29:42,004 rewrite it so it was good, and then submit that as a second poll request 464 00:29:42,155 --> 00:29:48,280 within minutes, which does not scale for crap, but it was so welcoming and so. 465 00:29:48,959 --> 00:29:51,389 You belong here, that it really led to an 466 00:29:51,389 --> 00:29:54,360 amazing community for a while that doesn't scale. 467 00:29:54,360 --> 00:29:56,790 And I'm not suggesting anyone should necessarily follow 468 00:29:56,790 --> 00:30:00,030 that pattern, but man, was I fortunate to encounter that, 469 00:30:00,425 --> 00:30:03,720 that, I mean, that was a great gesture on his behalf. 470 00:30:03,899 --> 00:30:12,209 Um, I feel like, you know, if, if I am, if I think that PI is, uh, 471 00:30:13,470 --> 00:30:18,449 addresses a real issue and uh, there's a good justification to add it. 472 00:30:18,885 --> 00:30:23,355 It's, it's very likely that I would do some rewriting to it or ask the, the, 473 00:30:23,415 --> 00:30:25,365 you can also, this was also a different era. 474 00:30:25,425 --> 00:30:27,255 You can assume good faith. 475 00:30:27,315 --> 00:30:31,425 Back in 2012 when I was doing these things, now for all, you 476 00:30:31,425 --> 00:30:34,755 know, someone just threw something into a slop cannon and 477 00:30:34,755 --> 00:30:37,755 basically said, I need to have another one for my resume. 478 00:30:37,785 --> 00:30:39,585 Go, go submit a pr. 479 00:30:39,615 --> 00:30:41,535 And they don't, they don't know what the project is. 480 00:30:41,535 --> 00:30:42,735 They don't care about it. 481 00:30:42,885 --> 00:30:46,545 At least historically, it used to be people going through and fixing the comma. 482 00:30:46,655 --> 00:30:49,024 Placement in your docks fine, whatever. 483 00:30:49,295 --> 00:30:50,915 Now it's a lot harder to tell. 484 00:30:50,975 --> 00:30:53,435 We actually open, uh, a few. 485 00:30:53,615 --> 00:30:57,065 We have a number of issues that we are opening to get help 486 00:30:57,065 --> 00:31:00,875 from the public and we're willing to pay for those prs. 487 00:31:01,595 --> 00:31:05,975 Oh, no, that, I'm sure this will not cause an in influx of AI generated prs. 488 00:31:06,035 --> 00:31:06,545 Oh no. 489 00:31:06,605 --> 00:31:09,725 Yeah, that's exactly what is happening these days. 490 00:31:09,754 --> 00:31:11,465 Uh, people try to know too. 491 00:31:12,195 --> 00:31:18,465 Address those PIs and collect, uh, that money by simply throwing an AI at it. 492 00:31:18,825 --> 00:31:22,095 And then we're, we're, you know, we're bombarded. 493 00:31:22,095 --> 00:31:23,085 We've seen crap like this. 494 00:31:23,085 --> 00:31:27,045 Like I get at this on my security email inbox all the time for last 495 00:31:27,045 --> 00:31:30,525 week in AWS, which is again, a podcast and an email newsletter. 496 00:31:30,764 --> 00:31:33,675 It is not a SaaS app or anything like it. 497 00:31:33,675 --> 00:31:36,135 And they're always finding the same things of you're not. 498 00:31:36,245 --> 00:31:40,655 Enforcing hard fail for SPF on your emails because if you do that, it will in 499 00:31:40,655 --> 00:31:44,104 fact cause some problems with email forwarding in some newsletter edge cases. 500 00:31:44,435 --> 00:31:46,145 So there you should fix that. 501 00:31:46,145 --> 00:31:46,985 Pay me please. 502 00:31:47,014 --> 00:31:48,604 Like, what the hell is this? 503 00:31:48,695 --> 00:31:50,254 It's the lowest effort. 504 00:31:50,254 --> 00:31:51,395 It's the beg bounty. 505 00:31:51,935 --> 00:31:52,445 It's awful. 506 00:31:52,564 --> 00:31:52,985 Awful. 507 00:31:53,074 --> 00:31:54,485 We're getting those as well. 508 00:31:54,784 --> 00:31:57,185 Um, I mean, at least twice a week. 509 00:31:57,185 --> 00:31:59,854 I'm getting an email of that sort. 510 00:32:00,570 --> 00:32:01,010 I don't know. 511 00:32:01,010 --> 00:32:02,330 It's a, it's a new era. 512 00:32:03,044 --> 00:32:03,615 I think 513 00:32:03,825 --> 00:32:06,375 it is, and I would not wanna be starting my career in this era. 514 00:32:06,375 --> 00:32:08,054 I, I don't have a better path for this. 515 00:32:08,060 --> 00:32:10,695 The, the roads you and I walked have long since closed. 516 00:32:10,695 --> 00:32:12,405 It's the world's a different place. 517 00:32:12,495 --> 00:32:16,575 I, I'm always scared to give Boomer tear advice of, well, back in my day, just 518 00:32:16,665 --> 00:32:20,145 you print out your resume, a nice paper, and you walk into the office and you 519 00:32:20,145 --> 00:32:23,175 demand to speak to the owner and you have a firm handshake and hit the bricks. 520 00:32:23,175 --> 00:32:24,435 You'll have a job by sundown. 521 00:32:24,585 --> 00:32:25,125 Yeah, sure. 522 00:32:25,125 --> 00:32:25,514 That'll work. 523 00:32:26,264 --> 00:32:27,645 Now they'll call the police. 524 00:32:27,705 --> 00:32:31,889 I see people, colleagues of mine or you know, family who. 525 00:32:32,620 --> 00:32:35,980 Who are working, uh, in the tech industry and the 526 00:32:35,980 --> 00:32:40,030 picture that they paint is, is, is pretty awful. 527 00:32:40,090 --> 00:32:47,260 I mean, from my perspective, people who, you know, call themselves poels, are 528 00:32:47,260 --> 00:32:52,719 not, are not typing anything in the keyboard except for handing, handing their, 529 00:32:52,810 --> 00:32:58,570 uh, work over to an AI and then just ing stuff in without even looking at it. 530 00:32:59,385 --> 00:33:02,625 And then you ask them, you know, okay, how does it work? 531 00:33:02,985 --> 00:33:04,845 Or maybe you can walk me through this. 532 00:33:04,905 --> 00:33:07,065 And the answer is, I have no idea. 533 00:33:07,545 --> 00:33:08,715 AI generated it. 534 00:33:09,705 --> 00:33:14,235 And so I think, I think that's a very dangerous route to take. 535 00:33:14,475 --> 00:33:15,765 I mean, there is the counterpoint too. 536 00:33:15,765 --> 00:33:17,355 I look at this thing like, how does this work? 537 00:33:17,355 --> 00:33:18,015 I don't know. 538 00:33:18,015 --> 00:33:19,725 I wrote it six months ago like that. 539 00:33:19,725 --> 00:33:21,105 That was always my answer too. 540 00:33:21,105 --> 00:33:22,995 It's like, honestly, I don't know what the hell me 541 00:33:22,995 --> 00:33:25,095 of six months ago was doing, but he was terrible. 542 00:33:26,385 --> 00:33:29,175 I can, I can relate to that as well. 543 00:33:29,410 --> 00:33:29,629 But 544 00:33:30,795 --> 00:33:33,615 how many times in your career have you been like, what idiot wrote this? 545 00:33:33,615 --> 00:33:36,015 And you fire up, get blame, and it turns out it was you. 546 00:33:36,015 --> 00:33:39,405 So you quietly fix it, don't say a word, and then just go ahead and patch it. 547 00:33:39,735 --> 00:33:40,155 Yeah. 548 00:33:40,215 --> 00:33:40,575 Yeah. 549 00:33:40,635 --> 00:33:45,945 It's fun to look back and see, you know, how you as an individual. 550 00:33:46,455 --> 00:33:49,395 Developed over, over the years, especially if you are involved with certain 551 00:33:49,395 --> 00:33:50,235 values of fun. 552 00:33:50,235 --> 00:33:50,775 Yes. 553 00:33:52,425 --> 00:33:56,685 Uh, I mean, if you are involved in the project for, uh, for 554 00:33:56,685 --> 00:34:00,885 a few years, then hopefully you, you do get to evolve and 555 00:34:01,545 --> 00:34:04,365 it's nice, you know, to have this last retrospect and see. 556 00:34:05,820 --> 00:34:08,580 How, how was the code that you, you have written 557 00:34:08,580 --> 00:34:11,310 five years ago and, and where you're at right now? 558 00:34:11,460 --> 00:34:15,719 Hopefully, hopefully you have improved, but, but, but these days, you 559 00:34:15,719 --> 00:34:23,219 know, people who are, uh, using ai, uh, to the point where they, they don't 560 00:34:23,219 --> 00:34:27,779 give it, they don't, don't even get the chance to practice their craft. 561 00:34:28,830 --> 00:34:31,020 Um, yeah, I think that's problematic. 562 00:34:31,020 --> 00:34:32,940 People are saying it is, you know. 563 00:34:33,360 --> 00:34:38,190 Um, in, in my area, people are saying that new developers, 564 00:34:38,190 --> 00:34:43,620 juniors are very, are having a really hard time, um, getting work. 565 00:34:44,040 --> 00:34:51,120 And so I, I, I wonder, uh, where, where would we find the, the seniors, the 566 00:34:52,290 --> 00:34:57,120 15 years ahead of us, the, the seniors from, where are they gonna come from? 567 00:34:57,675 --> 00:34:59,565 That's the thing that I've been pointing out to people. 568 00:34:59,565 --> 00:35:01,995 When I first found that AI could write some code and showed it 569 00:35:01,995 --> 00:35:04,905 to some friends, their responses, well, this is great to replace 570 00:35:04,905 --> 00:35:07,695 junior engineers, but it won't replace senior engineers up. 571 00:35:08,235 --> 00:35:10,515 Where, where do you think senior engineers come from, buddy? 572 00:35:10,515 --> 00:35:13,305 Uh, they don't spread early form from the forehead of some God somewhere. 573 00:35:13,365 --> 00:35:14,115 Exactly. 574 00:35:14,120 --> 00:35:17,985 And, and maybe some would argue that in the, you 575 00:35:17,985 --> 00:35:21,435 know, in the near future, seniors can also be. 576 00:35:21,779 --> 00:35:22,589 Replaced. 577 00:35:23,460 --> 00:35:24,419 I, I don't know. 578 00:35:24,480 --> 00:35:27,960 I think that at the moment LLMs are having 579 00:35:27,960 --> 00:35:32,009 difficulty to work with large code bases. 580 00:35:32,399 --> 00:35:37,589 Uh, I do see them as a great utility to start out something new. 581 00:35:38,009 --> 00:35:41,460 Whenever there's a clean slate and you ask it, okay, bill 582 00:35:41,460 --> 00:35:45,450 for me from scratch something, it might, you know, work 583 00:35:45,450 --> 00:35:50,129 well, it might work well for applications that are not. 584 00:35:51,120 --> 00:35:58,080 At the system level, um, maybe fronted front end type of work, maybe some 585 00:35:58,080 --> 00:36:07,140 backend type of work, but to build, I dunno, 20,000 line worth of, uh. 586 00:36:07,560 --> 00:36:12,990 Some high performance, high quality, uh, application database. 587 00:36:13,259 --> 00:36:13,379 Yeah. 588 00:36:13,379 --> 00:36:16,319 My version of UX is around ergonomics of command line tools. 589 00:36:16,319 --> 00:36:18,000 I am not a front end person. 590 00:36:18,000 --> 00:36:21,509 This, this has been an unlock for a lot of different use cases 591 00:36:21,509 --> 00:36:24,419 there, and it's powerful, but I'm also not sitting here thinking 592 00:36:24,419 --> 00:36:27,089 I can build a reasonable front end for anything of scale. 593 00:36:27,629 --> 00:36:30,330 The counterpoint is, I think that's a, that's a dangerous point. 594 00:36:30,330 --> 00:36:33,359 Like, well, that's not gonna scale at in any meaningful way. 595 00:36:34,185 --> 00:36:37,425 It's a back of house application for one email 596 00:36:37,425 --> 00:36:39,675 newsletter to get from one API to another. 597 00:36:39,675 --> 00:36:41,535 Maybe it doesn't have to scale. 598 00:36:41,715 --> 00:36:44,955 Maybe in the event that there's an outage in something, 599 00:36:45,075 --> 00:36:47,175 we can send the newsletter a half hour later. 600 00:36:47,175 --> 00:36:47,535 Maybe. 601 00:36:47,535 --> 00:36:48,675 It's not a concern. 602 00:36:48,915 --> 00:36:52,095 It's, it's fit the problem for business case, because by the 603 00:36:52,095 --> 00:36:54,795 time you go from concept that you're puttering around with 604 00:36:54,795 --> 00:36:57,525 one evening on your laptop to a hyperscale service, you're 605 00:36:57,525 --> 00:36:59,865 gonna have rewritten that thing a dozen times over already. 606 00:36:59,865 --> 00:37:00,285 Anyway, 607 00:37:00,525 --> 00:37:01,245 I, I agree. 608 00:37:01,815 --> 00:37:03,404 Like my favorite part, especially a startup founder 609 00:37:03,404 --> 00:37:06,045 myself, is every time you hire someone, they're better 610 00:37:06,045 --> 00:37:08,505 at the thing that you're hiring 'em for than you are. 611 00:37:08,715 --> 00:37:11,895 So it's a constant state of being humbled by these things 612 00:37:11,985 --> 00:37:14,145 and watching people in the interview trying to put a 613 00:37:14,145 --> 00:37:17,384 diplomatic corporate spin on what the hell is wrong with you? 614 00:37:17,384 --> 00:37:18,525 Why would you do it this way? 615 00:37:18,615 --> 00:37:23,085 Which, well, it got us this far to hire you, so please come help fix it. 616 00:37:24,555 --> 00:37:29,085 Uh, if, if people wanna learn more about Falco DV and graph 617 00:37:29,085 --> 00:37:31,875 databases at large, where's the best place for them to go? 618 00:37:31,965 --> 00:37:37,095 Well, they can check out our website, uh, they can check out our documentation. 619 00:37:37,335 --> 00:37:42,195 Uh, I dunno if, uh, we're gonna, uh, include the links to those or Well, we 620 00:37:42,195 --> 00:37:44,025 absolutely can put those in the show notes for you. 621 00:37:44,025 --> 00:37:44,805 It'll be fine. 622 00:37:44,865 --> 00:37:50,415 Then our documentation, our website and our disco channel. 623 00:37:50,475 --> 00:37:51,134 This is where. 624 00:37:51,720 --> 00:37:56,940 Uh, you can reach out and, uh, learn about, uh, our 625 00:37:56,944 --> 00:38:01,259 product or our service and on graph, uh, generally. 626 00:38:01,830 --> 00:38:05,580 And if you are, you know, interested in the actual source code 627 00:38:05,580 --> 00:38:09,240 and see seeing what I'm working on, then you can check out 628 00:38:09,240 --> 00:38:14,549 our GitHub, which is Falco db at falco db or slash falco db. 629 00:38:15,480 --> 00:38:17,520 All of that will wind up in the show notes. 630 00:38:17,720 --> 00:38:19,830 Roi, thank you for taking the time to speak with me. 631 00:38:19,830 --> 00:38:20,430 I appreciate it. 632 00:38:20,430 --> 00:38:20,495 Thank you, 633 00:38:20,504 --> 00:38:20,855 Colby. 634 00:38:21,600 --> 00:38:22,319 It was fun. 635 00:38:22,470 --> 00:38:27,000 Likewise, Roi Lipman, CTO, and co-founder of Falco db. 636 00:38:27,299 --> 00:38:30,690 I'm Cloud economist Corey Quinn, and this is Screaming in the Cloud. 637 00:38:30,990 --> 00:38:31,980 If you've enjoyed this podcast. 638 00:38:32,430 --> 00:38:35,609 Please leave a five star review on your podcast platform of choice. 639 00:38:35,640 --> 00:38:39,210 Whereas if you didn't enjoy this podcast, please leave a five star 640 00:38:39,210 --> 00:38:42,480 review on your podcast platform of choice, along with an angry, 641 00:38:42,480 --> 00:38:45,240 confused comment that makes it sound like you're a zookeeper. 642 00:38:45,240 --> 00:38:47,879 Understanding why giraffes need their own database.