[00:00:00] Degrees and regrets --- [00:00:00] Whitney Lee: So Kote, I think I've mentioned on the show before, maybe a few times, my [00:00:05] degree is in fine art, specifically in photography, which was, um, [00:00:10] over 20 years ago I earned this degree and I was thinking, I don't know [00:00:15] what your degree is in, what's your degree? [00:00:16] Coté: Oh boy. This is exciting. [00:00:20] My, my, my degree, uh, I, I went to, uh, the [00:00:25] University of Texas in Austin Uhhuh, and my degree is in, uh, philosophy. [00:00:28] Coté: And then I have a, [00:00:30] I have a, just needed to take one more class minor in English. Uh, which, [00:00:34] Whitney Lee: so you [00:00:35] don't have a minor in English is [00:00:36] Coté: what? No, no, I do. I dot I'm [00:00:37] Whitney Lee: kidding you. [00:00:38] Coté: Yeah. Yeah. It, it, it would've been, [00:00:39] Whitney Lee: oh, I [00:00:40] see low hanging fruits [00:00:42] Coté: minor. I have, I have a, I have a mid-size, uh, degree [00:00:45] in English, but I just needed, I had to take one more class and I was just like, I can't [00:00:50] stand this anymore. [00:00:51] Coté: And, uh, I don't know. This is one of those things as an adult, you're like. [00:00:55] Past self was an idiot. They should have just taken the seminar where a bunch of people [00:01:00] talked about some book they read and they would've had this credential. But at the time of school, yeah. [00:01:05] [00:01:05] Philosophy to tech path --- [00:01:05] Whitney Lee: Do you, so the question is, so what you do now, did you have [00:01:10] any idea what you might do as a career, as a philosophy major, or, or were you just like, this sounds [00:01:15] fun. [00:01:16] Coté: Yeah, I, I did it. 'cause, 'cause [00:01:20] uh, I, I, I liked the material because I had, uh, uh, I'd [00:01:25] started programming in 94, 95 when I was in high school. Uhhuh, [00:01:30] because I think I've told this story briefly, but, uh, one of the, the, the [00:01:35] fathers of a, a fellow student, a classmate that's what you call him, [00:01:40] was, had a startup to do, like, this is the nineties. [00:01:42] Coté: So they had a online mall and [00:01:45] they realized that, uh, high schoolers are very cheap. So you can carry them. So [00:01:50] I had a, I had a programming job all throughout there, so my degree was never really [00:01:55] career minded. It was just like writing [00:01:56] Whitney Lee: stuff, oh, you [00:01:57] Coté: kind of knew you were [00:01:58] Whitney Lee: gonna have a [00:01:58] Coté: tech [00:01:58] Whitney Lee: career [00:02:00] and then you, [00:02:01] Coté: I don't, that's giving me a little too much credit. [00:02:03] Coté: I didn't even think about that. It was [00:02:05] just like, I like studying this. I have a job. So it never really entered my mind to [00:02:10] get a degree that made me employable. [00:02:12] Whitney Lee: Oh, to be young again. [00:02:15] [00:02:15] Coté: I know. It's like an, an extremely lucky series of events [00:02:20] that I just sort of like stumbled into. So, so, [00:02:24] Liberal arts skills at work --- [00:02:24] Whitney Lee: so a follow up [00:02:25] question, even though you didn't intend it, do you think that your degree in philosophy has aided your tech [00:02:30] career in any way? [00:02:31] Coté: Uh, yeah. Yeah. I mean there's a lot of, I mean, this is a [00:02:35] typical Gen X or liberal arts position, right? It's just like, well now I can read and [00:02:40] study anything and then write about it, right? Like, and so, ah, that's a cliche way of [00:02:45] putting it. But yeah, I think for the work that I do, and even when I was. Only a programmer.[00:02:50] [00:02:50] Coté: Yeah. Like, it, it makes it, uh, it, it ma it makes it possible [00:02:55] that you can actually like, comprehend and communicate things. Yeah. And so I [00:03:00] think, and I, and I think the, uh, I think both in, in like the [00:03:05] o of course this is the type of English, I don't know if this is peculiar to American stuff, but you know, [00:03:10] you read books and you talk about their deep essence and stuff. [00:03:13] Coté: Uhhuh it wasn't creative writing or anything. [00:03:15] Uh, and, and that with like philosophy, it's for the role after I was a [00:03:20] programmer, being an analyst, and now whatever it is, I do like, it's, it's pretty, [00:03:25] it's pretty good for like analysis. Yeah. And then, and then the other thing that I think is important, and then, and then [00:03:30] I'll stop is like, I guess you would call it needing to [00:03:35] come up with a position whether you care or not. [00:03:37] Coté: Like, you know, one might, one might call it a [00:03:40] take, but it's sort of like when, when you study those fields. You can't really, [00:03:45] I mean, I tried several times, but you can't really write the paper. There's no paper to write here like [00:03:50] this. [00:03:50] Whitney Lee: I, I take issue with the fact that there always has to be a take about everything [00:03:55] in our modern society. [00:03:56] Whitney Lee: But, um, I don't wanna open that can of worms because I [00:04:00] want, I'm excited to introduce our, our guest already, [00:04:03] Coté: maybe in society. Yes. Uhhuh. [00:04:05] But just in the, in, in, in the, the milieu of work. I mean Yes. Yes. We don't need to [00:04:10] open the can worms. Right. [00:04:12] Whitney Lee: Sandwich versus Taco. Who cares? I don't [00:04:15] care. [00:04:15] Coté: Exactly. [00:04:17] Meet Josh Burka --- [00:04:17] Whitney Lee: Um, hello Josh. [00:04:19] Whitney Lee: Bur is [00:04:20] welcome to Hey there to the show. Will you introduce yourself please and tell us your college [00:04:25] degree? [00:04:25] Josh Berkus: Yeah. So I'm Josh Burka. Um, I, [00:04:30] I contribute to a variety of open source projects. Um, Kubernetes [00:04:35] being the currently the most prominent [00:04:37] Whitney Lee: Uhhuh, [00:04:38] Josh Berkus: um, have had a history, [00:04:40] um, working at various open source projects for the last.[00:04:45] [00:04:45] Josh Berkus: I used to say 25 years, but now it's more like 29, isn't it? Um, [00:04:49] Whitney Lee: time [00:04:49] Josh Berkus: [00:04:50] keeps doing [00:04:50] Whitney Lee: that. Yeah. [00:04:51] Josh Berkus: Yeah. Um, the, um, [00:04:55] uh, I work for Red Hat, um, and, um, I am, [00:05:00] uh, talking to you here from Portland, Oregon, so. [00:05:02] Whitney Lee: Awesome, awesome. [00:05:03] Josh Berkus: Um, I also have a [00:05:05] liberal arts degree. Um, [00:05:06] Whitney Lee: yeah, [00:05:07] Josh Berkus: I, yeah, uh, my major was fine art, [00:05:10] um, like you Oh, me too. [00:05:10] Josh Berkus: In my case. In, yeah, in sculpture, um, [00:05:13] Whitney Lee: sculpture. Cool. [00:05:14] Josh Berkus: So, [00:05:15] um, I still do pottery. [00:05:19] Whitney Lee: Oh, oh my [00:05:20] God. He's holding up a, a handmade Kubernetes, uh, mug. Yeah. Ceramic [00:05:25] mug. It's beautiful. [00:05:26] Josh Berkus: Yeah. [00:05:26] Coté: Yeah. That is [00:05:27] Josh Berkus: nice. Yeah, I made these for the Kubernetes one 11 [00:05:30] release team because I was released Oh, look at that. For one 11, so. [00:05:33] Whitney Lee: Wow. [00:05:34] Josh Berkus: Um, and so how [00:05:34] Whitney Lee: [00:05:35] of them did you make? [00:05:37] Josh Berkus: Uh, 40, I think. Um, [00:05:39] Whitney Lee: wow. [00:05:39] Josh Berkus: [00:05:40] It's, um, it, it's was customary for, um, the [00:05:45] release team, the release lead to give a gift to everybody who was on the [00:05:50] release team Uhhuh. Um, I, at the time that I did it, it was something you provided [00:05:55] yourself. Um, the release team has gotten a little bit too large, [00:06:00] um, and now the CNCF actually pays for it. [00:06:02] Josh Berkus: So, um, [00:06:05] but, um, and so, so I got the liberty in, in [00:06:10] fine art. Mm-hmm. Um, like, um, uh, coat, [00:06:15] I was, um. One, [00:06:20] I almost had a second major in [00:06:25] economics. [00:06:25] Whitney Lee: Okay. Do you [00:06:27] Josh Berkus: But [00:06:28] Whitney Lee: yes, [00:06:28] Josh Berkus: as far as economics was concerned, [00:06:30] it was still the Reagan years. Reagan. And so I was really Reagan interested [00:06:35] in conservative monetarist economics. [00:06:37] Josh Berkus: Um, fair. If, if I had been [00:06:40] four years younger, I actually would've gotten the degree in economics because economics changed a lot, um, in [00:06:45] that amount of time. [00:06:45] Coté: Oh yeah. [00:06:46] Josh Berkus: Um, the, um, uh, but at the time I was not interested [00:06:50] in, in monetary theory, uh, which seemed very boring to me. And, and kind of beside the point, [00:06:55] the, um, yeah, [00:06:57] Sculpture and learning to learn --- [00:06:57] Whitney Lee: I'm really interested in how or whether your [00:07:00] college degree, whether sculpture degree has informed your. [00:07:03] Whitney Lee: Your career [00:07:05] in tech? [00:07:05] Josh Berkus: Yeah. Well, leaving aside, you know, the stuff that [00:07:10] I've designed for actual projects, um, [00:07:13] Whitney Lee: uhhuh, [00:07:14] Josh Berkus: although not as much [00:07:15] because I did not study design. Um, and so if I could locate at somebody who's an actual [00:07:20] designer, I have them do stuff. Um, Uhhuh [00:07:22] Whitney Lee: absolutely. [00:07:22] Josh Berkus: The, um, um, I mean, you know [00:07:25] this, but a lot of our, you know, but a lot of people who, who don't have any experience in the arts don't realize that [00:07:30] design is actually separate from other divisions of, of art and has its [00:07:35] own, you know, sort of set of, of skills and focuses. [00:07:39] Josh Berkus: Um, and, [00:07:40] and if you're an artist, you can punt as a designer, but you're aware that you're not really a real designer. [00:07:45] Mm-hmm. Um, the, um, so [00:07:50] where, where the liberal arts degree was useful, um, is that a big part [00:07:55] of a liberal arts curriculum is learning how to learn stuff? Yeah. [00:08:00] And as I say to people. You know, my entire [00:08:05] field, um, you know, actually in both incarnations of my technical career, [00:08:10] right? [00:08:10] Josh Berkus: So first starting with databases and then going from there [00:08:15] to, um, cloud, right? [00:08:17] Whitney Lee: Uhhuh [00:08:17] Josh Berkus: did not exist when I graduated college. [00:08:20] [00:08:20] Whitney Lee: Oh. [00:08:21] Josh Berkus: Um, like I didn't study CS in college at all because at the time I [00:08:25] was going to school, CS was part of the math department, and [00:08:30] I'd already dropped outta physics because I realized I didn't like higher math. [00:08:34] Whitney Lee: Mm-hmm. [00:08:34] Josh Berkus: Um, [00:08:35] so, um, I wasn't interested in hanging around the math department. A lot of my friends were [00:08:40] in cs. Um, but I, but I wasn't for that reason. So I had this fun thing of, I [00:08:45] had an art degree that I paid for partly. By running the University [00:08:50] of Computer Lab. [00:08:51] Whitney Lee: Oh, big V cluster. So you had early, just like Kote you were [00:08:55] doing Yeah. [00:08:56] Whitney Lee: Tech stuff on [00:08:56] Josh Berkus: the, the side. Yeah. [00:08:57] Atari hacking origins --- [00:08:57] Josh Berkus: I started, I started when I was 13. Um, [00:09:00] Whitney Lee: Uhhuh, [00:09:00] Josh Berkus: uh, you know, this was when the eight bit PCs came out. Um, [00:09:05] and, um, Atari, the gaming company had their own PC for a [00:09:10] while. [00:09:11] Coté: Mm-hmm. [00:09:11] Josh Berkus: And so if you were interested in [00:09:15] video games, which would, what teenager wasn't? Um, this was the era of video game arcades, [00:09:20] right? [00:09:20] Josh Berkus: Um, um, it was an opportunity to [00:09:25] design your own video games. Um, and if you just [00:09:30] learned some programming, um, oh, you know, which I did starting at 13 years old. So, [00:09:35] um, and by the time I graduated I [00:09:40] was actually writing. Programs for my dad who was a [00:09:45] physician. I mean, here's the funny thing is my dad also started out in computers. [00:09:48] Josh Berkus: Uhhuh, he, he did the opposite way. Uhhuh. He [00:09:50] started out in computers in college, but this was during the vacuum tube era. [00:09:55] Um, and decided that he wanted a [00:10:00] job. He wanted to deal with people more. And you know, this was, 'cause he's a [00:10:05] boomer Uhhuh, he could go back to school for free. Um, and so he got a medical [00:10:10] degree and he's, um, uh, ended up in, in obgyn. [00:10:14] Josh Berkus: [00:10:15] So. [00:10:15] Whitney Lee: Oh, fascinating Mike experience [00:10:17] Coté: getting a biomechanic some. [00:10:20] [00:10:20] Josh Berkus: Yeah. [00:10:20] Coté: So, so you were, you were a, you were an Atari person then? I, I don't, mm-hmm. I don't think many people [00:10:25] who had Ataris. That's, that's a, you know, Ataris as pc. [00:10:27] Josh Berkus: It wasn't very successful. [00:10:30] It wasn't very successful indeed. But it was actually, it. [00:10:34] Josh Berkus: In my [00:10:35] opinion, um, the Atari was the best [00:10:40] of the APIC computers for learning. Um, because it was very [00:10:45] easy to take apart, um, compared to say Apples, the [00:10:50] computers were relatively affordable. And so you weren't afraid to take them apart. Yeah, [00:10:55] yeah, yeah. And there was a whole, the Atari developed a whole, um, [00:11:00] hacker culture around it. [00:11:02] Josh Berkus: Like, um, when, by [00:11:05] the time the Atari 400 expired and we all had eight hundreds or sts, um, we took one of [00:11:10] those old, like I joined, there was an Atari club in my town, Uhhuh, and I [00:11:15] joined the Atari Club. This is in San Antonio, Texas, actually Uhhuh. Um, and I joined the [00:11:20] Atari Club and we took an old 400, um, took a lot of it apart, [00:11:25] added new wiring, fused it with a CB radio, and made a primitive voice [00:11:30] synthesizer. [00:11:31] Josh Berkus: Um, [00:11:31] Whitney Lee: very cool. [00:11:33] Josh Berkus: So [00:11:33] Whitney Lee: that's awesome. [00:11:35] [00:11:35] Josh Berkus: Yeah. And if you look at that, that kind of led directly into a portion of my [00:11:40] career much later in, in database because the biggest thing, [00:11:45] the thing I got paid the most money for in databases was performance optimization. Right. [00:11:48] Whitney Lee: Uhhuh. [00:11:49] Josh Berkus: But I feel [00:11:50] like if you started out in those days where you were literally modifying your computer with a soldering iron [00:11:55] [00:11:55] Whitney Lee: Yeah. [00:11:55] Josh Berkus: You had a lot better understanding of the [00:12:00] parts of a PC and how they might affect performance [00:12:03] Whitney Lee: Yes. [00:12:03] Josh Berkus: Than [00:12:05] people who started out in the cloud, right? [00:12:06] Whitney Lee: Yeah. You get like, where is this literally big run? [00:12:10] What compute is lighting up with this? Like where is the memory? Right? Where is the network, the [00:12:15] actual literal wires connecting this to this? [00:12:17] Whitney Lee: Yes. [00:12:18] Josh Berkus: Right. And [00:12:18] Whitney Lee: that's cool. [00:12:20] [00:12:20] Josh Berkus: Yeah. And, and to think about that in terms of, of like, hey, maybe this is [00:12:25] not a software book, right. Like one of our, for a whole bunch of our clients, um. [00:12:30] You know, so you, this is now leaping ahead like [00:12:35] 15 years later, right? But we have the Postgres company, right. And a [00:12:40] whole series of our clients have crippling performance problems that are taking their websites down. [00:12:44] Josh Berkus: [00:12:45] And um, you know, and we're going through this, and this is before [00:12:50] everybody migrated to Amazon, so we could still check on what kind of hardware they had. [00:12:55] Um, and um, you know, this is when, well, everybody was on Rackspace actually. [00:13:00] Um Oh, [00:13:00] Whitney Lee: okay. [00:13:01] Josh Berkus: And what had happened was Rackspace had bought a new [00:13:05] generation of Dell equipment, and Dell had this whole generation of [00:13:10] release of their affordable servers that had a bad North Bridge.[00:13:15] [00:13:15] Whitney Lee: Oh. [00:13:15] Josh Berkus: Now for people who are not hardware people, the north is the [00:13:20] chip that shuttles data, it shuttles bits between the CPU and the [00:13:25] memory banks. So if that is not operating well. [00:13:30] The entire computer slows down to a crawl. And what would happen is once you went over a [00:13:35] certain amount of traffic over the North Bridge, it essentially shut down. [00:13:39] Josh Berkus: [00:13:40] Um, and, um, and I do kind of feel like [00:13:45] if I, and actually a couple of my, you know, um, colleagues, a couple of [00:13:50] my partners on the business hadn't come from that era where having a [00:13:55] computer meant monkeying with the hardware yourself, it would've taken us a lot longer to [00:14:00] identify that we were dealing with an actual hardware problem. [00:14:03] Hardware matters again --- [00:14:03] Coté: Um, do you, do you think, do you think that era of, [00:14:05] uh, caring is the wrong word, but getting a lot of benefit of knowing [00:14:10] about the hardware and the software working together, like, does that still exist? Like, you [00:14:15] know, and on one side, you know, you listen to the, uh, the oxide people talk and you're like, ah, it's alive [00:14:20] and well. [00:14:20] Coté: Like they're, they're, they're going crazy with the hardware and the software, but then they're the only ones. Like, [00:14:25] unless maybe it, I don't seem to encounter a lot of [00:14:30] conversation where I, I mean, I'm a software person, so this sounds odd, but where the hardware [00:14:35] matters, like, like where something's going on with it, it seems unseen nowadays. [00:14:39] Josh Berkus: [00:14:40] Well, but, so I would say the [00:14:45] problem with that is it still matters, right? And, and particularly we're seeing that now, right [00:14:50] now, that we're getting into the generation where people are building their own AI models and [00:14:55] AI agents and that sort of thing. Mm-hmm. Suddenly the hardware matters again. Right. You need to have the right [00:15:00] GPU, you need to have a bunch of these other things. [00:15:02] Josh Berkus: It matters how many CPU threads you [00:15:05] have, everything else. Right. Suddenly the hardware matters again. And we have, [00:15:10] we've had a couple generations of programmers that have gone their [00:15:15] entire software careers not thinking about hardware at all. Right? [00:15:18] Coté: Right. [00:15:18] Josh Berkus: And so [00:15:20] it's really hard. They're suddenly having to catch up here. [00:15:24] Josh Berkus: [00:15:25] Whereas like for example, weirdly like, [00:15:30] like, you know, we have the armed people coming in and, and you know, pushing a lot of multi [00:15:35] architecture and everything else, right? And you look at that and you're like, it's [00:15:40] mostly a bunch of older folks, um mm-hmm. Driving a [00:15:45] lot of that. And part of the reason is because older folks [00:15:50] already had the multi arch experience, right? [00:15:52] Josh Berkus: Mm-hmm. We had that multi arch experience back in the early aughts, [00:15:55] right? And so going back to multi arch is not this [00:16:00] mysterious, you know, unknown universe to us. [00:16:03] Whitney Lee: Mm-hmm. [00:16:04] Josh Berkus: But something we were [00:16:05] familiar with we're like, Hey, is X 86 starting to suck? Well, maybe we should actually go [00:16:10] to, you know, something else. [00:16:11] Josh Berkus: We try something else. Right? Those arm uhhuh, the new arm chips are looking kind of [00:16:15] cool. Let's try 'em out. Um, the, um. You know, [00:16:20] whereas if your entire career has been on top of X 86, you never thought about [00:16:25] what processor you were using before, um, then moving to [00:16:30] having multiple CPU architectures is very frightening. [00:16:33] Josh Berkus: You don't, you don't know what's [00:16:35] involved. Right. You don't know what you don't know. Right. [00:16:37] Whitney Lee: Yeah, exactly. You, you might not even [00:16:40] know. I mean, I am like younger in tech, a medium in tech, and, um, [00:16:45] and I, I don't, I wouldn't know to even think about the hardware level if I was having [00:16:50] that problem. So, yeah. E exactly. [00:16:52] Whitney Lee: You don't know what you don't know. How do you begin to [00:16:55] find that that's the problem as someone who's not introduced to the concepts? [00:17:00] That's rhetorical, but it's also real. Like, is it just Yeah. [00:17:05] Yeah. [00:17:06] Josh Berkus: Yeah. [00:17:06] Making 40 Kubernetes mugs --- [00:17:06] Coté: I have, I have, I have a unrelated question that I wanted [00:17:10] circle back to, and that is Yeah. So you, you made, you made 40 cups for the Kubernetes.[00:17:15] [00:17:16] Coté: Yeah. And, and like, so. You know, you're, you're trained in [00:17:20] this, you've got a certificate of, of knowledge and training. Uh, you know, and I know, I know you do, you [00:17:25] got a store and everything. Obviously you're good at doing it. Like, what do you do differently when you make [00:17:30] like 40 cups versus one cup? Like how, how does that, how does that get planned out and [00:17:35] handled differently? [00:17:35] Coté: Or is it exactly the same? You just make [00:17:37] Josh Berkus: one [00:17:37] Coté: cup 40 times? [00:17:39] Josh Berkus: Yeah. [00:17:40] Well, particularly, I mean, there is the one cup, right? So the body of the cup right, is going to be the same for [00:17:45] each. And, and by the way, because this was 11 one release, [00:17:50] um, this cup is particularly patterned after a cup used in, [00:17:55] um, the, um, the Tolkien movies, the, um, the, [00:18:00] uh, fellowship, the ring, et cetera, right? [00:18:02] Josh Berkus: Yeah. Right. They have a, they have a scene that [00:18:05] takes place in, um, a pub in the Shire and they have Uhhuh Mons that look like [00:18:10] this. Um, the, um, and um, [00:18:15] the. So, so the body is the same. [00:18:18] Production pottery lessons --- [00:18:18] Josh Berkus: I mean, the thing was, um, the ceramics I studied with like, uh, Paul s Soldner was the head of the ceramics department, right? So Paul s Soldner came from a tradition of production, pottery and, um, the Japanese method of, of learning crafting [00:18:40] Whitney Lee: uhhuh. [00:18:40] Josh Berkus: And, and both of those involve, uh, volume, [00:18:45] um, and repetition. Oh, right. Like one of our assignments, one of the terms, like the second term [00:18:50] you were in Paul's studio, for example, your assignment was make a hundred or [00:18:55] something. [00:18:57] Whitney Lee: Wow. [00:18:57] Josh Berkus: It doesn't care what, but you have to make a hundred [00:19:00] of them. Right? And, and up until then, you know, I'd been making, [00:19:05] you know, like maybe six, nine [00:19:10] vessels per term, right? [00:19:11] Josh Berkus: Mm-hmm. So it was a very different [00:19:15] experience, but the moment that you actually do that, you start thinking about, you know, first of all, you know, [00:19:20] how can I make this more consistent? Mm-hmm. Um, how can I make this faster? Right? Yeah. [00:19:25] Um, you know, um, the hundred, you could be trying to make a hundred identical [00:19:30] ones, you know, in which case, you know, you're actually sort of perfecting a shape or you can do what I do because I [00:19:35] was focused on sculpture, which is you made a hundred variations around a theme. [00:19:39] Josh Berkus: Oh. [00:19:40] Um, the, um, so, so playing with the same idea and doing [00:19:45] it a hundred different ways. Um, the, um, but either way it [00:19:50] changed how you think about, um, doing something right. [00:19:55] Because one of the biggest mistakes, going back to ceramics, is one of the biggest mistakes I see people who [00:20:00] are new to ceramics, particularly pottery in the wheel make, is they're throwing a vessel [00:20:05] on the wheel. [00:20:05] Josh Berkus: It goes badly. Mm-hmm. And they try to fix it. [00:20:09] Whitney Lee: Uh, [00:20:09] Josh Berkus: and so like, no, [00:20:10] don't fix it. Cut it off the wheel. Start over. [00:20:12] Whitney Lee: Yeah. [00:20:12] Josh Berkus: Because you'll spend an hour trying to [00:20:15] fix it and it'll still look terrible. Right, [00:20:18] Coté: right. [00:20:20] [00:20:20] Josh Berkus: The, um, so, um, the, [00:20:25] so that's part of it. But then, but then I really cut out my work for me because [00:20:30] I had each person's name embossed. [00:20:33] Josh Berkus: Oh [00:20:34] Whitney Lee: [00:20:35] goodness. I didn't even notice [00:20:36] Josh Berkus: that [00:20:36] Whitney Lee: part. [00:20:36] Josh Berkus: So, so I was actually making 40 different ones. [00:20:39] Whitney Lee: So [00:20:40] there's a Kubernetes logo embossed for those who can't see it, besides it being like [00:20:45] beautifully glazed with a different color on the inside and the outside. [00:20:48] Josh Berkus: Yeah. [00:20:48] Whitney Lee: There's Kubernetes logo [00:20:50] embossed, as well as a name embossed as well as the one do one, one, [00:20:55] uh, emboss too. [00:20:57] Josh Berkus: Yeah. Uh, by the way, the, the one 12 release lead really [00:21:00] hated me because she's like, I gonna do now. The, um, so, [00:21:05] um, the, um, so that meant I actually had to make 80 of them. [00:21:09] Coté: Huh? [00:21:10] [00:21:10] Josh Berkus: Because, because you have in any large, you know, if you have a kiln full of [00:21:15] wear, right? You're making 40 cups, right? A few of them, like even at my [00:21:20] level of experience, you can guarantee a few of them won't come out, right? [00:21:22] Josh Berkus: Mm-hmm. The glaze will bubble, the handle will [00:21:25] crack, something else will happen. And because of the [00:21:30] timeline, I didn't want to do some over from scratch, right? [00:21:35] Because those people would be getting their stuff five months later. At this point, I was actually [00:21:40] in my own studio, but it was when I could only access part-time. [00:21:44] Josh Berkus: [00:21:45] So from throwing to firing, it was like four to six months. [00:21:50] Today, um, I'm part of a co-op, and so [00:21:55] if I actually have the time to spend in the studio, which unfortunately my, my work travel schedule really [00:22:00] interferes with, um, the, um, you know, I can go from [00:22:05] throwing something to having a finished thing in about. [00:22:08] Josh Berkus: Three, four weeks. [00:22:10] Um, and that, [00:22:10] Coté: that's like one item, or can you just do a batch of them? Like, you could do that? [00:22:14] Josh Berkus: Well, no, you do a [00:22:15] batch, right? You do a batch, [00:22:15] Coté: right. Vessel. But, but you could [00:22:17] Josh Berkus: do, right. Yeah. [00:22:17] Coté: Could you do 10 vessels in that time [00:22:20] period? Or does it cap out at like 20? Or like what is the kind of like maximum? [00:22:23] Josh Berkus: What? It depends. It depends on how much free time I [00:22:25] have and importantly, how used to producing that thing I am. [00:22:29] Coté: Mm-hmm. [00:22:29] Josh Berkus: Because when I [00:22:30] produce a new shape, there's a lot of time that I spend figuring out how to [00:22:35] throw that shape consistently. Oh, right, right, right. Mm-hmm. So, um, [00:22:40] so for example, um, I had a thing a couple years ago where [00:22:45] I did one of the sort of pottery milestones for the first time. [00:22:49] Josh Berkus: You know, even though I've been doing [00:22:50] pottery for a really long time, I just never done this before, which is to produce a full set of dinnerware for somebody. [00:22:55] [00:22:55] Whitney Lee: Okay. [00:22:55] Josh Berkus: So ten four piece place settings. [00:22:59] Whitney Lee: Wow. [00:23:00] [00:23:00] Josh Berkus: So that's 40 individual pieces, right? Plates in two sizes, [00:23:05] bowls in two different shapes. And, um, and, and in that case I was [00:23:10] specifically copying something that they used to have. [00:23:13] Josh Berkus: Um, [00:23:15] so, um, so that required, it required the production [00:23:20] of some new tools for one thing uhhuh that, that I had to spend some time tinkering with. And then when I [00:23:25] did that, I actually had to experiment a lot with how exactly are these shapes thrown? Um, you know, [00:23:30] what's a, a good way to do it, right? Because like one of the things, one of the things that was the most painful is these wide past bowls [00:23:35] uhhuh. [00:23:35] Whitney Lee: And [00:23:35] Josh Berkus: the problem with the wide pastable is they really wide rims. And the problem with those wide rims is [00:23:40] when the clay is wet, they tend to fall [00:23:42] Whitney Lee: over. Oh, so, so is this [00:23:45] person who you're making, like, this seems like a tremendous amount of work. Is someone commissioning you [00:23:50] or is it someone who's really dear to you? [00:23:51] Whitney Lee: Yeah. [00:23:52] Josh Berkus: No, no, no, no, no. They just paid me for it. Um, [00:23:54] Whitney Lee: [00:23:55] okay. [00:23:55] Josh Berkus: The, um, so. They paid me for it. [00:24:00] Um, I mean, I didn't charge as much as somebody who's a professional full-time Potter would because [00:24:05] I knew it was gonna take me a long time, and they were okay with that, right? Mm-hmm. Like in the end, it took [00:24:10] me two years to actually complete their dinner set, um, [00:24:15] uh, because of the problems I ran into, because of some of the production problems and everything else. [00:24:19] Josh Berkus: Um, [00:24:20] the other thing is what I ran into was that my fellow co-op members uhhuh, so [00:24:25] plates take up a lot of room in the kiln. [00:24:26] Whitney Lee: Yeah. [00:24:27] Josh Berkus: Um, because in the glaze [00:24:30] firing, right, in the, in the bis firing, they don't, you can stack 'em, right? Mm-hmm. Glaze firing, you can't [00:24:35] stack them. And, and so anything that is wide takes up an inordinate amount of room in the [00:24:40] kiln, right? [00:24:40] Josh Berkus: Because, and so. My kiln [00:24:45] mates were only willing to let me fire so many plates at a time. [00:24:48] Whitney Lee: That makes sense. [00:24:50] That's, [00:24:50] Josh Berkus: um, [00:24:50] Whitney Lee: a lot of this kind of relates to software in terms of repeatability. Yeah. In terms of working [00:24:55] alongside other people with the same mm-hmm. Equipment. Um, can I take this opportunity to kind of ask, both of [00:25:00] you have been around tech for a long time Yeah. [00:25:02] Whitney Lee: And I've been around like six years maybe. And I, I [00:25:05] have some questions that I thought maybe you could answer. I guess a couple. Okay. [00:25:09] Sun Microsystems explained --- [00:25:09] Whitney Lee: One is, one is something that keeps coming up on this show is Sun Microsystems. What is Sun Microsystems and why is it so, why is everyone so touched it? [00:25:21] Josh Berkus: Yeah. So I used to joke, you know, back when [00:25:25] I was in Silicon Valley, right? [00:25:27] Josh Berkus: Mm-hmm. Um, so as, as somebody who was a [00:25:30] programmer, you know, and this is, I made this in the mid aughts, right? If you were a professional full-time [00:25:35] programmer in the mid aughts in Silicon Valley, [00:25:37] Whitney Lee: Uhhuh, [00:25:37] Josh Berkus: I used to joke that everybody had, at some [00:25:40] point in their career, worked for either some Microsystems or Cybase. [00:25:43] Josh Berkus: Um, [00:25:44] Whitney Lee: oh, I've never [00:25:45] heard of Cybase. [00:25:45] Josh Berkus: Yeah. Um, yeah, it's 'cause they're gone now. Um, okay. And, and they, they [00:25:50] actually got buried early in the aughts, um, for honestly very good reasons. You know, [00:25:55] like, um, sun, they were good at making software and terrible at running a business [00:26:00] and Yeah. Um, yeah, the, um, so, [00:26:05] um, both of those companies [00:26:08] Whitney Lee: uhhuh [00:26:08] Josh Berkus: churn through people, [00:26:10] right? [00:26:10] Josh Berkus: Like despite the fact that Sun existed as a company for whatever it was, 25 years, [00:26:15] something like that. Uhhuh, the average tenure of an engineer at Sun was two years. [00:26:20] And so over 25 years, if they're effectively rehiring all of their [00:26:25] engineers, every two years you go through like the entire programmer population of Silicon Valley. [00:26:29] Josh Berkus: [00:26:30] So [00:26:30] Whitney Lee: is it like AWS kind of known for that too? Are we not? [00:26:34] Josh Berkus: Um, [00:26:35] I think so except I, I think a lot more people. I mean, once a Ws, by the time [00:26:40] AWS came in the scene. The national and global programmer [00:26:45] population was much larger. [00:26:46] Whitney Lee: Ah, okay. That makes sense [00:26:48] Josh Berkus: because think about, we're talking about the [00:26:50] nineties, right? [00:26:51] Coté: I think AWS has a reputation of, since they have, uh, like an [00:26:55] accelerated vesting schedule over three years, pretty much everyone will stay there at least three years to get [00:27:00] the full payout. And then you sort of figure out like, I like this culture, or I don't, and it's [00:27:05] very black and white, as it were. Yeah. [00:27:06] Coté: And I think, I think people like, sort of figure [00:27:10] that out and they either stay or they go. [00:27:12] Josh Berkus: Yeah. [00:27:13] Whitney Lee: I, I wanna hear more about how [00:27:15] sun's terrible at running a business. [00:27:18] Josh Berkus: Oh. Um, [00:27:20] so I mean, the thing is, you had a lot of this in the nineties and Silicon Valley, [00:27:25] right? 'cause the world of, uh, sort of software and importantly internet exploded [00:27:30] uhhuh. [00:27:30] Josh Berkus: And as a result, there were a lot of companies that [00:27:35] had a good and very marketable technical idea. And they were able to make so [00:27:40] much money with that, that it didn't matter that the rest of their business skills were [00:27:45] non-existent. The, um, what did [00:27:48] Whitney Lee: Sun Microsystems [00:27:50] actually make? [00:27:52] Josh Berkus: Uh, servers. [00:27:53] Whitney Lee: Okay. [00:27:54] Josh Berkus: [00:27:55] So, so if you were hosting a large scale commercial [00:28:00] website, Uhhuh, like if you were running eBay, for example, to, to pick an, a [00:28:05] specific example that I happen to know about, if you were running like the eBay site in [00:28:10] 1998, which was already and had millions of users, et cetera, right. [00:28:14] Whitney Lee: Uhhuh, [00:28:14] Josh Berkus: those were [00:28:15] running on what we called sun pizza boxes. [00:28:17] Whitney Lee: Okay. [00:28:18] Josh Berkus: Um, uh, the Sun [00:28:20] Naturals, we call them pizza boxes because that was really the shape of the machines. Mm-hmm. Um, and you'd [00:28:25] have like big racks of these and um, they had, sun had their [00:28:30] own computer chip. Right. Um, um, and [00:28:34] Coté: operating [00:28:35] system. [00:28:35] Josh Berkus: Yeah. [00:28:35] Josh Berkus: And, and operating system. I was a BSD based operating [00:28:40] system, uh, one of the, the, uh, BSD branches. Um, and, um, like, [00:28:45] like at [00:28:45] Coté: the, that startup I was at, we had, we had a, uh, Solaris box. Yeah, that's, that was the name of their operating system. So [00:28:50] you'd call it a, um, a Solaris box. It's, it's similar to like, um, [00:28:53] Josh Berkus: yeah, [00:28:53] Coté: I, I mean it [00:28:55] a whole other scale, but it's similar to like, well, everyone's gonna buy one of these now and [00:29:00] before this is before you had cheaper, before, you know, Dell and HP and [00:29:05] the Super Micros and all these people figured out really cheap servers. [00:29:08] Coté: And before Linux was just [00:29:10] widely accepted, it was just like, well, we have this for the internet and that's it. [00:29:15] And so ev everyone would buy these boxes. [00:29:18] Josh Berkus: Yeah. And, and they were [00:29:20] good boxes, right? Like the actual technical production was, was amazing. Like, [00:29:25] um, I actually ran my first web business. [00:29:30] Um, from 2000 to 2003, 2004 ish [00:29:35] because, um, I and a partner realized [00:29:40] that even if the crash had happened, people still needed web hosting. [00:29:44] Josh Berkus: And [00:29:45] in 2000 after the crash, we were able to get Netra Netra for [00:29:50] 10 cents on the dollar from their original price, maybe less even. Actually, they were really cheap [00:29:55] because all of these first wave internet startups had gone outta business. [00:30:00] Um, and in some cases they had gone outta business so abruptly with so little planning for [00:30:05] shutdown that you could literally buy equipment from a former employee outta their garage. [00:30:09] Whitney Lee: Wow. [00:30:09] Josh Berkus: [00:30:10] Right. Um, so it was, it was something that was colloquially known at the time as five finger severance [00:30:15] because a lot of these employees had not been paid. Um, so they had not been paid, and so they [00:30:20] were basically making up their last paycheck by selling off the company equipment. So, [00:30:24] Sun culture and politics --- [00:30:24] Coté: and, [00:30:25] and, and, and, and tell me, tell me if, if, if my conception of it is wrong. [00:30:29] Coté: This part [00:30:30] was that, uh, well, sun had a, had a, had a, well you worked there, unlike me [00:30:35] just observing it so you would know better, but like, it was a very engineering driven culture, [00:30:40] which is to say there were engineers doing interesting things and trying stuff out, which is great. Yeah. You, you love [00:30:45] that. Now the, what's the word, uh, derisive? [00:30:49] Coté: The, the, [00:30:50] the, the negative judgy version of engineering led culture [00:30:55] means you make stuff that is not sellable. [00:30:57] Josh Berkus: Right. [00:30:58] Coté: And, [00:30:58] Josh Berkus: but that wasn't, [00:30:59] Coté: there's, [00:31:00] there [00:31:00] Josh Berkus: was a [00:31:00] Coté: fair amount of that going on. [00:31:02] Josh Berkus: Yeah. But that wasn't the main problem. [00:31:05] The problem is the leadership of Sun believed very strongly in social [00:31:10] Darwinism, which means that their way of cultivating new leadership was to [00:31:15] literally pit departments against each other. [00:31:17] Coté: Mm-hmm. [00:31:17] Josh Berkus: And we used to joke that [00:31:20] you could not get promoted to director unless you had stabbed at [00:31:25] least one other employee in the back. [00:31:27] Whitney Lee: Oh no. [00:31:29] Josh Berkus: And [00:31:30] this was a disaster for actually running a [00:31:35] business. I mean, to give you an example of, of one of these that I was part of, right. [00:31:40] [00:31:40] Postgres at Sun and MySQL --- [00:31:40] Josh Berkus: So I son briefly had a database department, um, that I was a manager. [00:31:45] Josh Berkus: And that was the reason why I went over was because I'd been working on Postgres. I was Postgres core team [00:31:50] for a while. And one of the things that I decided that Postgres needed more than anything in the world, [00:31:55] um, for spreading it all over and getting people to adopt it, was to have a tier [00:32:00] one vendor that actually supported it. [00:32:01] Josh Berkus: Right. And so when Sun approached me and they said, [00:32:05] Hey, we're putting together a database department. Postgres is gonna be one of our products, we'd like you to run it. I took it [00:32:10] up. [00:32:10] Whitney Lee: Mm-hmm. [00:32:11] Josh Berkus: And, um, the, um, [00:32:15] and I went, um. And [00:32:20] so, you know, we had the database department and, but obviously as a new [00:32:25] department and his son previously not having done their own databases and also having a partnership with Oracle, [00:32:30] it was difficult to build a business, right? [00:32:33] Coté: Mm-hmm. [00:32:34] Josh Berkus: Yeah. And so then we [00:32:35] had this sort of breakthrough. We ran into a customer that wanted to, you know, have this huge migration and that [00:32:40] sort of thing. They also wanted to buy a whole ton of storage and Sun had its own storage at the time. [00:32:45] Um, and um, and so it [00:32:50] was this million, million and a half dollar deal of [00:32:55] software in storage, et cetera. [00:32:56] Whitney Lee: Uhhuh, [00:32:58] Josh Berkus: well, [00:33:00] you know, we were part of as database, we were part of the Solaris department, [00:33:04] Whitney Lee: [00:33:05] Uhhuh, [00:33:05] Josh Berkus: so the head of the storage department. Pulled out of the deal just to [00:33:10] screw over the head of the Solaris department. [00:33:12] Whitney Lee: Oh [00:33:12] Josh Berkus: no. And the whole thing collapsed. [00:33:14] Whitney Lee: [00:33:15] And this [00:33:15] Josh Berkus: is, and that sort of thing happened constantly. [00:33:17] Josh Berkus: Yeah, [00:33:17] Whitney Lee: yeah. [00:33:19] Josh Berkus: Yep. [00:33:20] [00:33:20] Whitney Lee: That's wild. [00:33:21] Josh Berkus: Mm-hmm. [00:33:22] Whitney Lee: Uh, like why I don't get it. 'cause [00:33:25] it's, it only hurts everyone. [00:33:27] Josh Berkus: Yep. But, but it hurt the head of the [00:33:30] Solaris department more and I think they actually got replaced after that. [00:33:33] Whitney Lee: So, [00:33:33] Josh Berkus: and [00:33:33] Whitney Lee: there, which [00:33:33] Josh Berkus: makes, that was what they cared about. [00:33:35] Right. Which makes, [00:33:35] Whitney Lee: is that we ended up that person the most likely candidate for whatever promotion, if they can [00:33:40] knock out the other person. [00:33:41] Whitney Lee: It [00:33:42] Coté: is an organization ends up like that. It, you know. Yeah. It [00:33:45] like, there, it, it is, I, I haven't thought about it in a while, but they, [00:33:50] it, it seems like you kind of, you kind of hit on this, they Sun was also on this, um, [00:33:55] often deadly path, I don't know, portfolio [00:34:00] strategy, where you're doing not only the hardware, but also the software that you run on top of it.[00:34:05] [00:34:05] Coté: So like, and there's very few companies, maybe there's almost none that can pull [00:34:10] this off. But for example, you're saying like, as I recall, a lot of Solaris was moved [00:34:15] by Oracle and vice versa, right? Yeah. Like you, and, and so like you don't wanna piss off [00:34:20] Oracle and then, right. And then, and then here's Postgres. [00:34:23] Coté: Trying, trying to start up a business. And so, [00:34:24] Josh Berkus: well [00:34:25] the reason [00:34:25] Coté: it's, [00:34:25] Josh Berkus: yeah, it's [00:34:26] Coté: sort But the reason that they, a lot of conflict there. [00:34:27] Josh Berkus: Yeah. Yeah. The reason they fired [00:34:30] up our department, that was originally actually a good strategy because, well, I thought it was good strategy. [00:34:35] Lemme go later on. What actually was happening. [00:34:37] Josh Berkus: I thought it was good strategy because [00:34:40] the Sun Oracle partnership was in jeopardy. And so it was a [00:34:45] useful threat from Sun to say, Hey Oracle, you better renew the [00:34:50] partnership because otherwise we're gonna go into competition with you. [00:34:54] Whitney Lee: And [00:34:55] what was actually happening, [00:34:56] Josh Berkus: what was actually happening was that they wanted MySQL. [00:34:59] Whitney Lee: [00:35:00] Ah. [00:35:01] Josh Berkus: So, um, the Postgres, my SQL [00:35:05] rivalry from the early days was famous, but the truth is, in terms of the people who actually developed the [00:35:10] software, um, uh, we knew each other, some of us [00:35:15] were friends. We had a couple of major contributors who went back and forth between the two databases [00:35:20] because, um, you know, once again, going back to the nineties, early [00:35:25] aughts, you could have easily fit everybody doing actual database engineering into one room. [00:35:29] Josh Berkus: [00:35:30] Mm-hmm. And it wouldn't even have been that big of a room. So, um, [00:35:35] the, um, and so it was actually [00:35:40] a, if you look at it from a very smart, long-term strategy, but, [00:35:45] um, not something that once I figured it out, I left sun. Right, [00:35:49] Coté: right, right. [00:35:50] Mm-hmm. [00:35:50] Sun Plays Postgres vs MySQL --- [00:35:50] Coté: You, you wanna be a, uh, [00:35:52] Josh Berkus: yeah, [00:35:52] Coté: you, you, you don't wanna be, let's, let's call it a [00:35:55] potential. [00:35:55] Coté: Winner that ends up being a pawn. Right, right. That [00:36:00] [00:36:00] Josh Berkus: and, and yeah, and Sun hired me and started the Postgres [00:36:05] strategy both as an enticement and a threat to my SQL because [00:36:10] Sun was trying to buy my SQ l [00:36:12] Whitney Lee: Oh. [00:36:13] Josh Berkus: And, and they [00:36:15] knew both. I was very friendly with a lot of the, my SQL staff. Right. Um, [00:36:20] and also that, you know, once again they could say, Hey, my [00:36:25] sql, we're a tier one vendor, we can build up Postgres. [00:36:28] Josh Berkus: 'cause at the time, right, [00:36:30] the, my SQ l user base was easily 10 times the size of the Postgres user base. [00:36:34] Coté: Mm-hmm. [00:36:34] Josh Berkus: Right. Yeah. [00:36:35] Maybe 20 times. Um, the, um, and [00:36:40] but tier one vendor sponsorship would change that. It, it really would change that. Um, [00:36:45] as it turns out, other things change that on their own. But the, um.[00:36:50] [00:36:51] MySQL Deal and Sun Exit Plan --- [00:36:51] Josh Berkus: And Sun was buying my SQL not as a [00:36:55] business strategy, but because the executives were already planning to sell the company, so sell [00:37:00] Sun, and they knew that owning my SQL would drive would [00:37:05] cause both IBM and Oracle to raise their offers. [00:37:08] Coté: And, and do you, do you, uh, do you know how much [00:37:10] they paid from my SQL Whitney? [00:37:12] Coté: No. It, I, I believe it was $1 [00:37:15] billion, so something like that. Which, which I, which was [00:37:18] Whitney Lee: in the nineties. Oh my God. Yeah. [00:37:19] Coté: Right, right. [00:37:20] But, but I, but I bring up for comparison. It's like, oh [00:37:22] Josh Berkus: yeah. [00:37:22] Coté: That sounds like a Your monthly bill [00:37:25] [00:37:25] Josh Berkus: Yeah. [00:37:25] Coté: For you. [00:37:26] Josh Berkus: No, it was, it was 2008. But, but it's important to understand [00:37:30] that it was the most that had been paid for an open source company, like at this time.[00:37:35] [00:37:35] Josh Berkus: Absolutely. Mm-hmm. [00:37:35] Whitney Lee: Right. [00:37:36] Josh Berkus: Red Hat's revenue was just barely a billion dollars. [00:37:40] Right. So it was the most that had been paid for an open source company. [00:37:45] On record, you know, and pretty high up in terms of [00:37:50] the amount that was paid for any kind of a startup, right? I think PayPal was actually a sort of similar amount.[00:37:55] [00:37:55] Coté: Hmm. [00:37:55] Josh Berkus: Um, and PayPal had a huge revenue. So the um, [00:38:00] and um, I mean, and that was also part of a, a standard, you know, sort [00:38:05] of stock market resolution thing. Because if you look in going into the late aughts sun had a [00:38:10] problem that you don't wanna get into when you're running a company, which is that [00:38:15] Sun's total stock value had actually dipped [00:38:20] below their cash on hand. [00:38:22] Whitney Lee: Oh. [00:38:23] Josh Berkus: And you don't wanna be in that [00:38:25] position because then you are a target for corporate raters. Because if they can do a leverage buyout, then [00:38:30] they can use your cash on hand to pay off the buyout. [00:38:33] Whitney Lee: Oh, wow. [00:38:35] I've learned so much about business as a tech person. [00:38:38] Coté: Yeah. And then optimization city after [00:38:40] that. [00:38:40] Coté: It's, it's thrilling time. [00:38:42] Whitney Lee: Yeah. [00:38:42] When Big Companies Fail Upward --- [00:38:42] Whitney Lee: So do you think, so we've, you've talked about like quite a few ways now. Sun is, was doing business badly. Do you think they're like, um, modern day companies that, do you think modern day companies can get away with being bad at business now or at least in small ways? Or do you think that's just, it's too cutthroat [00:38:59] Josh Berkus: [00:39:00] now? [00:39:00] Josh Berkus: Yeah. [00:39:00] Whitney Lee: Yeah. I mean, [00:39:01] Josh Berkus: I'll, yes, and, and plenty of them are right, [00:39:05] because I look, look at it this way, and this is actually a mistake people do when they're looking at [00:39:10] businesses to emulate, right? [00:39:12] Whitney Lee: Mm-hmm. [00:39:12] Josh Berkus: Um, there's a tendency that when you look at a business that is [00:39:15] being financially successful, to assume that every single thing that business does is financially [00:39:20] successful. [00:39:21] Josh Berkus: Mm, good point. It can sometimes be the case that there's literally only one thing that that [00:39:25] business is doing that's financially successful, but that thing earns so much [00:39:30] money. That it makes up for everything else they're failing at [00:39:33] Whitney Lee: Uhhuh. [00:39:34] Coté: Yeah. [00:39:35] [00:39:35] Josh Berkus: Right. And, and I'll give you a couple of famous examples that other people have written about. [00:39:39] Josh Berkus: Right. [00:39:40] Um, you know, one is, um, Google is [00:39:45] notoriously bad at creating application platforms. [00:39:48] Whitney Lee: Oh, [00:39:48] Josh Berkus: right. [00:39:48] Whitney Lee: Uhhuh, [00:39:49] Josh Berkus: um, other [00:39:50] people have written about this, right, is that they just can't seem to hit on a formula that developers actually [00:39:55] want to adopt. Um, uh, for a variety of reasons, right? I'll let the people who [00:40:00] actually do application platforms who've written about this check out their words. [00:40:03] Josh Berkus: Right. I'm just saying this, [00:40:05] that, you know, they, they, there was this Google apps universe thing. [00:40:10] Um, they've had other generations of stuff. Um, Google Cloud has [00:40:15] had, you know, issues with developer uptake and that sort of thing. Um, [00:40:20] but Google's other businesses, you know, advertising, um, [00:40:25] search. Um, I, a whole [00:40:30] bunch of other things are successful enough Yeah. [00:40:32] Josh Berkus: That, that division of their business [00:40:35] can lose money more or less continuously. Right? [00:40:37] Whitney Lee: Yeah. Um, the, it's [00:40:40] not, I, I understand, I don't really know firsthand, but I understand it, it is a [00:40:45] good business strategy to have r and d in areas that you're growing in, or that, like you might [00:40:50] fail at first as long as you succeed eventually. [00:40:52] Whitney Lee: So I don't think, uh, some failure is [00:40:55] necessarily a problem in terms of business strategy, but it's more, [00:41:00] you're saying someone looking out from the outside to the inside at the business, getting the wrong idea [00:41:05] about how it's profitable. Yeah. [00:41:07] Josh Berkus: Yeah. Because, because like one of the things you might say is, [00:41:10] Hey, Google's a big successful company, right? [00:41:12] Whitney Lee: Uhhuh, [00:41:13] Josh Berkus: um, [00:41:15] I'm launching a new cloud host company. I'm going to emulate [00:41:20] Google Cloud's entire strategy. Right? [00:41:22] Whitney Lee: Okay. [00:41:22] Josh Berkus: You know? [00:41:23] Whitney Lee: Yes. [00:41:23] Josh Berkus: Ignoring the fact that that is the [00:41:25] least successful thing. Um, that, that Google has been doing historically [00:41:30] compared to all of their other businesses, right? [00:41:31] Whitney Lee: Mm-hmm. [00:41:32] Josh Berkus: Um, the, um, [00:41:35] so, um, you know, and, [00:41:40] and, and I'd say that's actually true of big businesses in general with big businesses in general, you would say, if [00:41:45] you look at the majority of things they're doing, the majority of things they're doing are either indifferent or not [00:41:50] successful. [00:41:50] Josh Berkus: It's just they have a few things that they're really good at and they're making a lot of money, and those make up for the rest. [00:41:55] Mm-hmm. Now, you know, super companies with super good strategy [00:42:00] and good centralized management start dropping the things that they're not good at. They start being realistic [00:42:05] about it, right? [00:42:05] Coté: Mm-hmm. [00:42:06] Josh Berkus: Um, but there is a tendency to keep trying to [00:42:10] say, oh no, if we restart this and to not realize that it might actually be a problem [00:42:15] with your company culture, right? Yeah. It might be a problem with your company culture that you just [00:42:20] cannot focus properly on this particular thing and you should stop trying to do it. [00:42:24] Josh Berkus: Um, [00:42:25] [00:42:25] Coté: like, like, like an example, one of the things, at least in my head that you're talking about there is Yeah. You look at, [00:42:30] uh, I worked there for a couple of years. You look at Dell and they, they're pretty good at [00:42:34] Josh Berkus: Yeah. [00:42:35] [00:42:35] Coté: Exploring new things and then dumping them when they're like, we're, we're good at, at [00:42:40] hardware, right? [00:42:41] Coté: Yes. So they, they spun up a whole business of services and then [00:42:45] I, I was there trying to help out with strategy for like, so a software group when they bought, uh, [00:42:50] quest and Cloud and everything. And then, you know, eventually that all kind of like was pushed down and [00:42:55] they, you know, they, they, they just have a, um, it's, it's a, it's it's interesting case of [00:43:00] how to try new things without completely tanking the business. [00:43:03] Coté: Right. And then, and then as you [00:43:05] pointed out, like divesting of things once you're, you're kind of done with them. Yeah. Which is, but not cool for [00:43:10] the divested, but it sort of like maintains the long-term viability of the company. [00:43:14] Josh Berkus: Yeah. [00:43:15] [00:43:15] Coté: Or yeah. I mean, mind you, some contrast, you could, you could look at like the, uh, the metaverse [00:43:20] over at, at Facebook land where it's sort of like. [00:43:24] Coté: That was a [00:43:25] big leap into completely unrelated stuff and with [00:43:30] lots of, with lots of commentary about, well, here's how it's totally related to cat pictures [00:43:35] or, or, you know, whatever goes on in Facebook at the moment. Yeah. But really it was just like way out of the, the [00:43:40] capability zone and extremely expensive. [00:43:43] Josh Berkus: Yeah. [00:43:44] VC Pressure and Copycat Startups --- [00:43:44] Josh Berkus: Um, but it, you know, and, and that's actually a good example, but the Metaverse is a good example of doing something in order to entice investors and not because you think it's gonna be successful. [00:43:55] Whitney Lee: Um, that's a good in thing that's happens in modern business that maybe I, I don't [00:44:00] think what's around so much in the sudden Microsystems era, which is just, uh, the VC [00:44:05] money that's now piped into everything and, and, um, trying to get, [00:44:10] trying to please investors rather than trying to please a potential user base. [00:44:14] Whitney Lee: Yeah. [00:44:14] Josh Berkus: [00:44:15] I, I mean, as an engineer I find where it's gotten [00:44:20] to, to be really a problem. In, in terms of, of the quality of what we're producing these days. [00:44:25] Because if you go back to the s right, which, you know, and, and you know, [00:44:30] even the nineties in the first year of internet, um, if you're running a startup, it was a choice. [00:44:34] Josh Berkus: Actually, [00:44:35] there were a lot of successful startups that did not take vc. [00:44:37] Whitney Lee: Mm-hmm. [00:44:39] Josh Berkus: [00:44:40] Um, or that effectively when we took minority vc, right. They took some [00:44:45] investment, but they never gave over control of the company, et cetera. Um, [00:44:50] and, um, and some that never took any at all. Um, [00:44:55] uh, and the, um, and that meant that, you [00:45:00] know, founders were looking a lot more, well hopefully, you know, looking a lot more at, [00:45:05] is this investment really going to help grow the business or not, right? [00:45:09] Whitney Lee: Mm-hmm. [00:45:09] Josh Berkus: [00:45:10] Rather than assuming they automatically had to take it. And, but then back [00:45:15] then, the amount of money that you got through VC was an order of magnitude, a couple orders of magnitude smaller. [00:45:20] [00:45:20] Whitney Lee: Yeah. [00:45:21] Josh Berkus: Um, the problem that we have at now is there is so much VC [00:45:25] money that you don't have a choice anymore, right? [00:45:28] Whitney Lee: Mm-hmm. [00:45:29] Josh Berkus: You [00:45:30] come up with a product idea, right? Your initial product idea for your company, you launch it. If [00:45:35] you get a bunch of press and you're not taking vc, what will happen is someone [00:45:40] else will launch a copycat idea, you know, [00:45:42] Whitney Lee: take vc, [00:45:43] Josh Berkus: you know, $50 million [00:45:45] worth of VC and bury you in marketing, and then they will buy [00:45:50] your company so that they'll actually have a good product. [00:45:52] Coté: How, how, how closely tied do you [00:45:55] think? I'm gonna speak in general terms here? Yeah. Like the, hmm, [00:46:00] the, the, the sort of happiness and health of the open source community [00:46:05] is tied to, like the software funding model. And what I mean by software funding model. Yeah. I'm [00:46:10] gonna call it a cycle instead of like, you know, uh.[00:46:15] [00:46:15] Coté: May maybe, you know, you, you, uh, you, you start a company in a garage, you get a little bit of [00:46:20] investing. You have an IPO and then, you know, you're the next IBMI mean, you know, that's kind of like [00:46:25] a, a traditional model. And uh, and then, and then there's a little bit of variation [00:46:30] where like, you know, with a little bit of sun stuff and nowadays it's kind of like, well, what if we never [00:46:35] IPO Right, right. [00:46:37] Coté: What, what if we just stay private forever and ever and ever [00:46:40] and we amassed this huge ball of money and we might acquire companies. [00:46:45] I mean, you know, there, there's, I think, I think through all three of ours careers 'cause things have changed [00:46:50] a lot in recent years. Like the kind of funding cycle funding for software has changed several [00:46:55] times. [00:46:55] Coté: And so how, like, closely tied is the, the open source [00:47:00] body, if you will, to that? [00:47:03] Open Source Shifts to Vendors --- [00:47:03] Josh Berkus: Well, [00:47:05] I mean, I would say that this happened earlier, but [00:47:10] I would say that we've. Due to economic pressures [00:47:15] that, you know, I don't, there was no way to work around. We've lost one [00:47:20] major thing out of open source, which is if you turn the clock all the way back to the [00:47:25] nineties, the majority of open source software was actually produced by end [00:47:30] users. [00:47:32] Whitney Lee: Oh. [00:47:33] Josh Berkus: Um, [00:47:35] and there were huge advantages to that because number [00:47:40] one, it meant that what was being produced was a lot closer to what people wanted. [00:47:45] Right. The, um, because they were the users. Right. [00:47:50] And the second part of it is there [00:47:55] wasn't this competition between, Hey, I want this to be free and available to everybody, to, [00:48:00] Hey, I need to make some money off of it. [00:48:01] Coté: Mm-hmm. [00:48:02] Whitney Lee: Mm-hmm. ' [00:48:02] Josh Berkus: cause you didn't need to make any money off of it. As a matter of your more [00:48:05] focus was, can my company save money by using this? Which [00:48:10] didn't require you to charge anybody else anything? Um, [00:48:14] Whitney Lee: what, why does one [00:48:15] model need to replace the other? Why couldn't they coexist? [00:48:18] Josh Berkus: Well, so what [00:48:20] happened was right between, between [00:48:25] 1995 and 2005, right? [00:48:28] Whitney Lee: Uhhuh, [00:48:30] [00:48:30] Josh Berkus: um, let alone, you know, 2010, 2014, whatever, um, [00:48:35] what happened was it became possible to make money off of [00:48:40] open source software, right? People figured out, hey, this can be freely available, but we can still make money, [00:48:45] right? [00:48:45] Coté: Mm-hmm. [00:48:46] Josh Berkus: Um, uh, my current employer, red Hat, kind of led the way on [00:48:50] that. Mm-hmm. Um, there were other redheaded contemporaries, but none of them, I don't know [00:48:55] any of them that did particularly well. [00:48:57] Josh Berkus: At that open source model. But [00:49:00] you know, once, once Red Hat became a billion dollar company, everybody else was like, Hey, there's [00:49:05] something here. Um, and um, [00:49:10] and the thing is, once it was possible financially possible to [00:49:15] become an open source software vendor, you [00:49:20] have this order of magnitude problem, right? [00:49:22] Josh Berkus: Mm-hmm. Which is no end user [00:49:25] engineer is gonna be able to spend more than a fraction of their time [00:49:30] on writing software. [00:49:31] Whitney Lee: Mm-hmm. [00:49:32] Coté: Let alone like, [00:49:32] Josh Berkus: whereas if you're an engineer working [00:49:35] for a vendor, you're spending full time, if you're working in a startup, you're spending double full time [00:49:40] Right. Working on producing software. [00:49:42] Josh Berkus: And so even though the end user is still [00:49:45] in the mix, their percentage of who's working on what shrinks down [00:49:50] to five, 10%. Um. But [00:49:55] the problem that that introduces for open source, right, is now that it belongs to vendors, [00:50:00] there needs to be a revenue model. [00:50:05] And sometimes that revenue model conflicts with open source [00:50:10] community needs. [00:50:12] Whitney Lee: Yeah. And so [00:50:15] then also as you know, open source is often trying to solicit [00:50:20] contributors. Like I kind of get the older model. If I'm a contributor on a, a [00:50:25] project that I love and use, that feels cool. But if I'm a contributor donating my time [00:50:30] to a project that's making another company money, that doesn't feel good.[00:50:35] [00:50:35] Whitney Lee: So how does that translate in, into the, the [00:50:40] experience of contributing to open source? Yeah. [00:50:43] Foundations and Neutral Governance --- [00:50:43] Josh Berkus: Well that's why we have open source [00:50:45] foundations now [00:50:46] Whitney Lee: Uhhuh, [00:50:46] Josh Berkus: right? Because that is a genuine, major problem. [00:50:50] And that's why, uh, the Linux Foundation exists. [00:50:54] Whitney Lee: But how does the [00:50:55] Linux Foundation help solve the problem? [00:50:57] Josh Berkus: So at [00:51:00] instead of, right, so, um, you know, [00:51:05] we take, I can talk about my employer in this regard, right? [00:51:07] Josh Berkus: We have a project, right? Mm-hmm. Um, [00:51:10] like say Pod Man, right? Uhhuh, which has gotten really popular, [00:51:15] [00:51:15] Whitney Lee: which [00:51:15] Josh Berkus: is a container runtime, um, used a bunch, uh, yeah, yeah. Okay. It's, it's, [00:51:20] um, uh, the way to run containers on machines, right? [00:51:23] Whitney Lee: Mm-hmm. [00:51:24] Josh Berkus: And, and also build [00:51:25] them and, and also, you know, shuttle them around, uh, working, you know, sort of [00:51:30] together with Kubernetes, et cetera, right? [00:51:32] Josh Berkus: And, um, and also these days a way [00:51:35] to build and run operating systems. But that's a, uh, a side thing. Yes. So, [00:51:40] um, the, um, um, you [00:51:45] know, but as long as it was a completely Red Hat owned [00:51:50] project. Contributions from other vendors, were going to be limited. [00:51:55] Right? Yep. So Red Hat donated it to the Cloud Native Computing [00:52:00] Foundation. [00:52:01] Whitney Lee: Mm-hmm. [00:52:01] Josh Berkus: So that other vendors like Microsoft [00:52:05] for example, um, or, um, you know, apple, [00:52:10] um, would feel that they would not have [00:52:15] second thoughts about putting somebody on it if they were using it. [00:52:19] Whitney Lee: Yes. [00:52:19] Josh Berkus: Um, [00:52:20] [00:52:21] Whitney Lee: it's still, it still doesn't, um, address why like an [00:52:25] indi an individual would wanna donate their time to open source anymore. [00:52:28] Whitney Lee: Yeah. But [00:52:30] it doesn't. Yeah. [00:52:31] Josh Berkus: Yeah. [00:52:32] Why Individuals Still Contribute --- [00:52:32] Josh Berkus: Well, I mean, there's plenty of reasons for individuals to do it. Like, one of the things I don't wanna get away from particularly, 'cause like these days I'm actually mentoring a lot of Indian contributors, [00:52:40] Whitney Lee: Uhhuh, um, [00:52:42] Josh Berkus: and. You know, and if we're looking at [00:52:45] individual programmers, don't forget, and this is, goes back to, you know, many years in open source, there's a career [00:52:50] motivation. [00:52:50] Whitney Lee: Yes. [00:52:51] Josh Berkus: Right. Which is, is if you do a lot of open source contributing, that [00:52:55] becomes your resume. [00:52:56] Whitney Lee: Mm-hmm. [00:52:57] Josh Berkus: Um, and, and like literally just two [00:53:00] days ago, I had one of my mentees send me a message that they [00:53:05] got their promotion and part of it was on the strength of their open source contribution record. You [00:53:10] know, [00:53:10] Whitney Lee: that's [00:53:10] Josh Berkus: wonderful. [00:53:10] Josh Berkus: Um, and, and I have another one waiting in my inbox where I'm supposed to write a letter of [00:53:15] recommendation for somebody at another company. Based on their open source contributions. [00:53:20] So, and, and if your leadership in an open source project, you should never [00:53:25] overlook the fact that whether or not you personally do, um, people have those career [00:53:30] motivations. [00:53:31] Josh Berkus: And you should cultivate those, right? Because a good reason to spend time on your [00:53:35] project. Um, and, and so you should look at that of, to say, Hey, how can [00:53:40] my individual contributors advance their careers through my project? Because [00:53:45] it's a win-win, right? Mm-hmm. They get promoted and you get contributions. [00:53:50] [00:53:50] Joining Kubernetes and Release Team --- [00:53:50] Whitney Lee: So you said up top that you work a lot with the Kubernetes project. [00:53:54] Whitney Lee: What, what's [00:53:55] your role within Kubernetes? [00:53:56] Josh Berkus: Um, right now it's, it's [00:54:00] kind of a grab bag of things. Uhhuh, um, the, um. [00:54:05] You know, mostly because like some of the other people who've been, I'm sort of second [00:54:10] generation with Kubernetes. Right? Okay. So I'm not part of the founding generation, but I was [00:54:15] recruited shortly after that [00:54:17] Whitney Lee: Uhhuh, [00:54:18] Josh Berkus: um, you know, mostly 'cause I was [00:54:20] already involved with Docker. [00:54:21] Josh Berkus: Um, a bunch of the early Kubernetes people were former [00:54:25] MySQL people. [00:54:25] Whitney Lee: Oh. [00:54:26] Josh Berkus: Oh. And, and so as a result they specifically reached out to me. [00:54:30] Um, the, um, and um, [00:54:35] the, so, um, you know, and, and [00:54:40] for me, you know, I was actually looking to solve certain other problems. Like I wasn't from, I [00:54:45] started out from end user perspective, right. [00:54:46] Josh Berkus: Which is I was managing customers with [00:54:50] a lot of software and managing the software installations across their [00:54:55] fleet of servers was becoming a real problem. Uh huh. [00:55:00] And vSphere, which was the largest solution at the time, was too expensive for a lot of my [00:55:05] clients. [00:55:06] Whitney Lee: Ah, [00:55:06] Josh Berkus: you know, the plus they were looking at needing to hire a [00:55:10] full-time vSphere admin. [00:55:11] Josh Berkus: Mm-hmm. Um, or actually pay us, because our company had a couple people who were [00:55:15] certified vSphere admins because it was a good service to rent out was very lucrative. Mm-hmm. Um, [00:55:20] but, but they didn't want to pay those, those fees, and so we were looking for something more affordable for them. [00:55:25] Um, the, um, so um, [00:55:30] so I came in, um, the, um, [00:55:35] and somehow got recruited very early on to [00:55:40] the release team. [00:55:41] Josh Berkus: Okay. Um, that was first being [00:55:45] formed. Right. Because initially all of the Kubernetes releases were done completely by Google [00:55:50] staff. Ah, um, the, um, [00:55:55] and um, and this was a problem, right? Because, [00:56:00] um, nobody including Google wanted Kubernetes to be an exclusive [00:56:05] Google project. But if you have to have Google employee rights to do a software [00:56:10] release, then that means that Google could never pull its engineers off of doing releases. [00:56:14] Whitney Lee: [00:56:15] Yes. [00:56:15] Josh Berkus: Um, and, and some of the individual engineers who are doing Kubernetes releases [00:56:20] wanted to personally move on to other things. [00:56:21] Whitney Lee: Okay. [00:56:22] Josh Berkus: So, [00:56:23] Whitney Lee: so you were not at Google [00:56:25] and got recruited? [00:56:25] Josh Berkus: I was not at situation. I was working for a Red Hat at this point already [00:56:29] Whitney Lee: Uhhuh. [00:56:29] Josh Berkus: So [00:56:30] Red Hat had their own like early container platform called Atomic, [00:56:35] um mm-hmm. [00:56:36] Josh Berkus: That I was hired, um, to, to work [00:56:40] on, um, and promote. Okay. And the, um, [00:56:45] but I was already dabbling with Kubernetes and stuff. Um, [00:56:48] Whitney Lee: okay. [00:56:49] Josh Berkus: The, as a matter of [00:56:50] fact, I joke that I'm, I'm probably the only one who actually had a production install of [00:56:55] Kubernetes on version 0.4. Um, [00:56:57] Whitney Lee: wow. [00:56:58] Josh Berkus: The, um, [00:57:00] it was a very peculiar use case, [00:57:03] Whitney Lee: Uhhuh, [00:57:03] Josh Berkus: um, for [00:57:05] which, weirdly, that early sort of beta ish version of Kubernetes worked better than I [00:57:10] think later production versions would've worked. [00:57:12] Josh Berkus: Wow. Um, the, um, so, [00:57:15] but, um, the, um, you know, so, [00:57:20] um, you know, I'm in there, I've been contributing to docs and that sort of [00:57:25] thing. Um, I didn't actually start to learn Golan at all until I got involved with [00:57:30] Kubernetes, because previously I'd been. Um, I attributed to Python. And by the way, going back to the liberal [00:57:35] arts degree thing, right, Uhhuh, it's one of the reasons why this learning how to learn thing is so important is if you [00:57:40] stay in computing, you end up having to shift programming languages multiple times, right? [00:57:44] Whitney Lee: Oh yeah. Uhhuh. [00:57:45] [00:57:45] Josh Berkus: Um, I started with basic, I learned COBOL back in the [00:57:50] nineties and, uh, Fortran, um, the, um, [00:57:55] which was actually useful. Um, I did, I joke, I did COBOL as late as 2014. Um, [00:58:00] wow. 'cause it's still around the, um, and um, [00:58:05] you know, then went on to, uh, Pascal [00:58:10] mm-hmm. And Visual Basic. And, uh, IC [00:58:15] and Pearl, um, and then Python and [00:58:20] Ruby. [00:58:21] Whitney Lee: Wow. [00:58:21] Josh Berkus: Um, yeah. And because, because, you know, [00:58:25] there's a cycle and at a certain point if you wanna keep working on new stuff and not just maintaining [00:58:30] old stuff, you have to ship programming languages. Absolutely. And so you just need to get used to the idea that, hey, [00:58:35] every five years I'm gonna be learning a new programming language. [00:58:37] Whitney Lee: The only constant is change. [00:58:40] [00:58:40] Josh Berkus: Yeah. And the, um, you know, it's the same thing. People are like, oh, you doing anything with [00:58:45] rust? I said, you know, I've seen stuff that Rus look cool, I just don't have a use case right now. Right. [00:58:48] Whitney Lee: Uhhuh, a [00:58:49] Josh Berkus: [00:58:50] use case will probably eventually come along and then I'll learn rust. [00:58:53] Josh Berkus: Right. The, um, [00:58:55] [00:58:55] Whitney Lee: so bringing, bringing it back to Kubernetes, you were on the release team at first and whether [00:58:59] Josh Berkus: [00:59:00] because mm-hmm. I was looking for a way to contribute, you know, with again, where I, I was not competent enough [00:59:05] with go to be contributing to actual core Kubernetes. Right. Uhhuh, um. [00:59:10] The, um, but you know, looking at the sort of areas of the project that needed help [00:59:15] mm-hmm. [00:59:15] Josh Berkus: We're going like, oh, we're forming this release team and everything else. And I'm like, well, you know, [00:59:20] I did a bunch of the PostgreSQL software releases. Um, I did software releases for open office. [00:59:25] Um, so I'm like, I've released experience, I'll raise my hand. No volunteer. [00:59:30] Um, and this was for version 1 0 8, I [00:59:35] think [00:59:35] Whitney Lee: Uhhuh. [00:59:36] Josh Berkus: Um, and this is the first time that each of the Google people was doing [00:59:40] releases took sort of an apprentice that we call a shadow. Um, and [00:59:45] so I was supposed to be shadowing somebody on marshaling all of the prs [00:59:50] that were going into the current release. Right. Wow. So maybe figuring out what should be tagged for the [00:59:55] release and what shouldn't be tagged for the release and everything else. [00:59:56] Josh Berkus: Mm-hmm. Except that the Googler, who I [01:00:00] was supposed to be shadowing then left Google and left Kubernetes. [01:00:04] Whitney Lee: Oh. [01:00:04] Josh Berkus: [01:00:05] So I ended up actually going right into the fire immediately to being in charge of marshaling all the [01:00:10] prs uhhuh because it was one of those things that actually did not require Google employee permissions to do. [01:00:14] Whitney Lee: [01:00:15] Okay. [01:00:15] Josh Berkus: Um, the, um, so, [01:00:20] so kind of straight into it, um [01:00:22] Whitney Lee: mm-hmm. [01:00:22] Josh Berkus: The, um, I, but [01:00:25] in the process, I and the other people leading those sort of early form release [01:00:30] teams came up with this entire release team system, um, [01:00:35] that is now sort of a development machine for Kubernetes. [01:00:38] Whitney Lee: Oh, wow. [01:00:38] Josh Berkus: Um, the, [01:00:40] um, where, you know, every release every four months has a system of leads and [01:00:45] shadows and some of the shadows get promoted to being leads for the next cycle, et cetera.[01:00:50] [01:00:50] Josh Berkus: And there's a whole defined system for how people are recruited and trained to work on the next [01:00:55] release. Uh, [01:00:55] Whitney Lee: that's impressive that it's, that all of Kubernetes actually [01:01:00] is impressive to me because isn't there something like 80. Or 90,000 [01:01:05] contributors to Kubernetes. [01:01:07] Josh Berkus: Yeah. But I wouldn't say those are not [01:01:10] actually all, I mean that's like over time. [01:01:12] Josh Berkus: Right? [01:01:12] Whitney Lee: Over, yeah. Like [01:01:14] Josh Berkus: how [01:01:14] Whitney Lee: many active [01:01:15] [01:01:15] Josh Berkus: contributors do you have? You'll discover that there's like, you know, 70,000 people who contributed one PE one, [01:01:20] you know, p right. [01:01:21] Whitney Lee: That's me. [01:01:21] Josh Berkus: The, um, so yeah. Um, I would say, [01:01:24] Whitney Lee: but [01:01:25] still managing [01:01:25] Josh Berkus: 10,000 [01:01:26] Whitney Lee: people [01:01:26] Josh Berkus: would say. Right? Yeah. Active, active now, you know, doing our whole [01:01:30] curve from the, the one patch contributors to the, the full-timers. [01:01:32] Josh Berkus: Right. [01:01:33] Whitney Lee: Uhhuh, [01:01:33] Josh Berkus: it's still, [01:01:35] let's see, according to our org member [01:01:40] records, seven or 800. [01:01:42] Whitney Lee: Okay. Yeah. [01:01:43] Josh Berkus: It's actually huge as far as open source [01:01:45] projects go. Right. [01:01:45] Whitney Lee: Absolutely. [01:01:47] Scaling Kubernetes with Process --- [01:01:47] Josh Berkus: The um, um, and it's [01:01:50] big enough that we need a management system. Um [01:01:53] Whitney Lee: mm-hmm. [01:01:54] Josh Berkus: [01:01:55] And at the time that we developed a lot of our systems, we had some advantages, [01:02:00] right? [01:02:00] Whitney Lee: Mm-hmm. [01:02:00] Josh Berkus: That I'm not sure whether or not other open source projects will ever have. Again, [01:02:05] it was certainly the first time in history that I'd ever seen this through open source. [01:02:09] Whitney Lee: Mm-hmm. [01:02:09] Josh Berkus: Which [01:02:10] was in the early days of Kubernetes, we just had all of the money that we [01:02:15] wanted. Right? [01:02:16] Whitney Lee: Ah, [01:02:16] Josh Berkus: it was the hot thing. [01:02:18] Josh Berkus: Companies, multiple companies were [01:02:20] throwing a lot of money at it, [01:02:21] Whitney Lee: Uhhuh. [01:02:21] Josh Berkus: And so we not only had what [01:02:25] really amounted to an excess of engineers Right. Contributing to it. Um, [01:02:30] so back in those days, in terms of the people who were like, there was probably over a thousand people. [01:02:35] Who were, you know, at least one quarter time on Kubernetes. [01:02:39] Josh Berkus: Right. [01:02:39] Whitney Lee: [01:02:40] Wow. [01:02:40] Josh Berkus: So just tons of input. Um, and importantly, this was also [01:02:45] the sort of golden days of open source practice offices at vendors. [01:02:50] And so we also had an entire cluster of [01:02:55] people who were management types, either community management or [01:03:00] product management, um, who were available to spend [01:03:05] time doing process development and doing organizational development for the project.[01:03:10] [01:03:10] Josh Berkus: Right. Ah, and um, and that was really important because it meant [01:03:15] that there were all of these processes we sent up, we set up. Right. [01:03:18] Whitney Lee: That's been able [01:03:18] Josh Berkus: to scale. Yeah. Um, because [01:03:20] yeah. 'cause we had people who had experience in developing processes [01:03:25] and who had time to think about the process for its own sake rather than, I need to fix this bug.[01:03:30] [01:03:31] Josh Berkus: Um, the, um, they, they had time to think about what is the bug [01:03:35] fixing process, right? [01:03:36] Coté: Mm-hmm. [01:03:36] Josh Berkus: How many people should sign off on it? How should the tests run? Right? [01:03:40] And, and that meant that we had, we have really good systems for it that [01:03:45] continue to pay off, especially now that we have less staffing than we used to have. [01:03:48] Josh Berkus: Right? Right. Now we have [01:03:50] that less staffing. We can actually rely on all the systems that were set up when we had more [01:03:55] people. [01:03:55] Whitney Lee: I understand. [01:03:56] Contributor Experience and AI Slop --- [01:03:56] Whitney Lee: Recently you've been, um, you were co-chair of SIG developer experience, so SIG stands for special interest group. [01:04:02] Josh Berkus: Yeah. [01:04:03] Whitney Lee: So, is that correct? That was a pretty Yeah. [01:04:05] Contributor [01:04:05] Josh Berkus: experience. Contributor [01:04:06] Whitney Lee: experience. Contributor experience. Ah, okay. What, what does that, [01:04:10] what are your duties as part of that special interest [01:04:12] Josh Berkus: group? Yep. So syn contributors [01:04:15] experience, speaking of that, that sort of management layer, right? Um, [01:04:20] oversees a bunch of stuff around what is it to be a contributor to [01:04:25] Kubernetes. [01:04:26] Josh Berkus: Right? [01:04:26] Whitney Lee: Mm-hmm. [01:04:27] Josh Berkus: And so this includes things like mentorship [01:04:30] programs, like ah, managing the list, managing GitHub [01:04:35] rights, managing the list of who is a current contributor. 'cause there's this concept [01:04:40] called org member, which is, you know, who's a current contributor and therefore, like on a technical level, is [01:04:45] allowed to interact with the CI bots. [01:04:47] Josh Berkus: Okay. [01:04:48] Whitney Lee: Ah, [01:04:49] Josh Berkus: [01:04:50] the, um, and, um, I, and who [01:04:55] manages the contributor website that has all the information around how to contribute, who runs? Mm-hmm. [01:05:00] Um, you know, contributor onboarding. Um, we have this thing called, [01:05:05] uh, the new contributor, uh, orientation. That makes [01:05:09] Whitney Lee: sense. Um, [01:05:09] Josh Berkus: [01:05:10] we've had a number of different programs for contributor onboarding. [01:05:12] Josh Berkus: Right. Uhhuh, um, like I started [01:05:15] out, uh, uh, Gwen and I and Paris started out [01:05:20] with a live one. Mainly it was GN and me started out with a live one at Coup Con. [01:05:25] Seven or eight years ago, um, uh, and [01:05:30] that, you know, worked for two or three years and then it, it, you know, requiring it to be in [01:05:35] person was too restrictive and we weren't getting the right people, and so they moved to an online version.[01:05:40] [01:05:40] Josh Berkus: Um, and, and I mentioned the NCO because, um, [01:05:45] the next, um, I believe that the next new contributor orientation is next week. [01:05:50] Um, [01:05:50] Whitney Lee: oh, exciting. [01:05:50] Josh Berkus: For anybody listening who's interested in contributing to Kubernetes, so look up Kubernetes new [01:05:55] contributor orientation. It happens two time slots, 12 hours apart so that people around the world can [01:06:00] join. [01:06:00] Josh Berkus: Um, [01:06:01] Coté: well that [01:06:02] Josh Berkus: nice, the, um, and, and so [01:06:05] in one way it's a grab bag of things, but another way it's like a, hey, somebody [01:06:10] needs to oversee, right? If you're gonna have 800 active contributors, right? [01:06:14] Whitney Lee: Uhhuh, [01:06:14] Josh Berkus: somebody [01:06:15] actually, literally somebody needs to be the HR department for those contributors. And that's contributor [01:06:20] experience. [01:06:20] Josh Berkus: That's contributor experience. Oh, of course. Um, the, um, and [01:06:25] um, and we co-founded that [01:06:30] sig, right? Uh, contributor experience was originally Paris Pitman's idea. [01:06:35] And so she co-founded that with several people, including me, to run it initially and build all these [01:06:40] systems, right? [01:06:40] Whitney Lee: Cool. [01:06:41] Josh Berkus: Um, and um, [01:06:45] and you know, then I was in the first generation and, you know, as the contributor management, we [01:06:50] really believed in leadership cycling, right? [01:06:52] Coté: Mm-hmm. [01:06:52] Josh Berkus: Um, you know, so after a [01:06:55] couple of years, you know, I recruited my replacement and I signed off from being lead of SIG contributor [01:07:00] experience. Went to work on other things. Um, I came back [01:07:05] because we had a little bit of a crisis where, you know, there's only, there was only [01:07:10] three leads for the sig and two outta the three leads for the sig had career [01:07:15] crises where they had to step down abruptly. [01:07:17] Josh Berkus: Not having recruited anyone to follow them. [01:07:20] [01:07:20] Whitney Lee: Ah. [01:07:20] Josh Berkus: Um, and so it's something that happens. One of the nice things about having this huge [01:07:25] pool of people who have been contributing for a while, right? This happens all the time, is if you actually have a leadership [01:07:30] crisis, you have this sort of emeritus pool that you can draw from, where you can say, Hey, [01:07:35] you know, Tim or whoever, can you just step in for long enough to recruit [01:07:40] somebody new to replace you? [01:07:41] Coté: You have some resiliency built into it, right? [01:07:44] Josh Berkus: Yeah. [01:07:44] Coté: Yeah. That [01:07:45] is, this seems pretty well staffed for that. [01:07:48] Josh Berkus: Yeah. [01:07:50] And, and like I said, and I didn't actually think about how important it was to have all these systems until [01:07:55] we're getting now into the age of AI assisted contributions. [01:07:59] Whitney Lee: Yeah. [01:07:59] Josh Berkus: [01:08:00] Right. And we're talking among projects about, um, [01:08:05] low quality ai. [01:08:06] Josh Berkus: Contributions [01:08:07] Whitney Lee: AI slot as they say. Yeah, [01:08:09] Josh Berkus: yeah. [01:08:10] Yeah. Because you know, the advantage to a lot of the AI assistance is that used, [01:08:15] well, it helps contributors improve the quality of their contributions. [01:08:18] Whitney Lee: Mm-hmm. [01:08:18] Josh Berkus: Used badly. [01:08:20] It allows contributors to theoretical to produce a higher volume of, of [01:08:25] just terrible contributions. [01:08:26] Josh Berkus: Yeah. Um, the, um, and, [01:08:30] and we were looking at how this was affecting Kubernetes and the answer was actually not [01:08:35] much. [01:08:36] Whitney Lee: Oh, that's [01:08:36] Josh Berkus: surprising. 'cause, because the truth was Kubernetes [01:08:40] having been a very popular major target for new people getting involved in open [01:08:45] source, you know, going back 10 years, right? [01:08:48] Whitney Lee: Mm-hmm. [01:08:48] Josh Berkus: We already developed systems [01:08:50] for dealing with high volume, low quality contributions. Right. [01:08:54] Whitney Lee: That's amazing. [01:08:54] Josh Berkus: [01:08:55] Right. From your, and so dealing with the AI assisted ones was not different. Yeah. [01:09:00] Um. The, um, and, and so like [01:09:05] there've been some specific things, right? Like things around reporting [01:09:10] CVEs that have been annoying, but they haven't actually risen to the level of, of annoying, uh, [01:09:15] beyond annoying, right? [01:09:15] Josh Berkus: And then when I talk to people from projects who are really struggling with this, it's [01:09:20] because they never had systems to deal with low quality contributions to begin with, right? [01:09:25] Mm-hmm. Um, they didn't have any way to evaluate contributions, you know, that [01:09:30] wasn't extremely labor intensive. [01:09:32] Whitney Lee: They didn't have a quantity problem yet, sot [01:09:34] Josh Berkus: Right, right. [01:09:34] Josh Berkus: [01:09:35] They didn't have a quantity problem yet, so they didn't have, they didn't have the concept of triage. Um, [01:09:40] they didn't have, you know, automated pre-submit tests. Um mm-hmm. [01:09:45] You know, they didn't have any of the other things that allows you to screen out a lot of stuff that's just not worth [01:09:50] looking at. Um, the, um, and, [01:09:55] um. [01:09:58] Josh Berkus: The struggle is real, right? Because [01:10:00] if you're on a project that only has four maintainers and like, [01:10:05] only one of those maintainers is actually really full-time on the project. [01:10:08] Whitney Lee: Yeah. [01:10:08] Josh Berkus: Um, [01:10:10] trying to develop those systems while continuing to patch stuff and get releases out and [01:10:15] everything else, it's really hard. [01:10:17] How to Get Involved and Wrap Up --- [01:10:17] Coté: Well, to to that point, why, why don't you tell [01:10:20] people again, like if they want to, uh, figure out how to be a contributor and help out with all of that, you have you [01:10:25] where, where, where do they go to sign up for this? Uh, the, the, the, [01:10:28] Josh Berkus: yeah, so look [01:10:28] Coté: at seminars, but, but the [01:10:30] things you were just [01:10:30] Josh Berkus: mentioning. Yeah. [01:10:30] Josh Berkus: Yeah. I mean, for, for Kubernetes looking for new contributor orientation. Right. [01:10:35] Um, I, I don't, that would be kind of the long route to help other [01:10:40] people's projects simply because yes, if you actually work on Kubernetes for a while and work with our systems and [01:10:45] everything else, you'll get a better idea of what you need to build for other projects. [01:10:49] Josh Berkus: But that [01:10:50] might be kind of the long way to do it, right? Mm-hmm. There's a lot of systems and it's gonna take you a [01:10:55] couple of years to get really sort of a grip on it. And some of the things you just [01:11:00] can't copy for a smaller project because like, it makes sense for Kubernetes to have its own 60 node [01:11:05] Kubernetes cluster for running CI software, right? [01:11:08] Whitney Lee: Mm-hmm. [01:11:08] Josh Berkus: It doesn't make sense for [01:11:10] a lot of smaller projects to do that. That's right. And, and things [01:11:15] like p pr, which is our custom built CI bot, is not really designed to work [01:11:20] to run on much smaller hardware installations. [01:11:25] [01:11:26] Coté: Yeah. And then, uh, and, and then, and then before we wrap up, like how about yourself? [01:11:28] Coté: So if people wanna see more [01:11:30] of what, uh, what you've been up to. Yeah. Do you have some, some place on the worldwide web that, that people [01:11:34] Josh Berkus: out? Yeah, you can [01:11:35] go, you can go to beku.org, B-E-R-K-U s.org. [01:11:40] Um, the, um, I, which has some links to, I'm [01:11:45] not, I'm not great about keeping it updated. Um, it mostly has links to old talks that I did.[01:11:50] [01:11:50] Josh Berkus: Um, the, um, but other places you can find me, I'm one of the [01:11:55] organizers for Portland, uh, cloud Native Kubernetes meetup. [01:12:00] Um, we have meetups once every two months. Um, also for DevOps meetup, which we [01:12:05] alternate with every other month. Um, the, um, [01:12:10] and, um, the, [01:12:15] um, and, um, I'm on the program committee for Southern California [01:12:20] Linux Expo, so you can find me there. [01:12:22] Josh Berkus: And at least [01:12:25] for the time being, my job involves me being at, at [01:12:30] least, uh, three Kons a year. [01:12:33] Coté: All over the place. [01:12:34] Josh Berkus: Yeah, all over the [01:12:35] place. So that's also a good place to find me. And of course on Kubernetes Slack and, and cloud [01:12:40] native Slack, um, also CAS Foundation. Um, the, um, [01:12:45] wow. So I, I'm not, it's not defined. [01:12:47] Josh Berkus: Yeah. I'm not a defined, and, and [01:12:50] one weird advantage that I got that you could never imagine is, [01:12:55] um, the fact that my family name is unique and [01:13:00] it's not a big family [01:13:01] Whitney Lee: Uhhuh. [01:13:02] Josh Berkus: So as a result it makes it easy for [01:13:05] people to find me on the internet. 'cause they type in Josh Burkes and they get me. There's be [01:13:09] Coté: five [01:13:10] different Josh Bur where Well, well, thanks for being on. [01:13:12] Coté: I, I think it, it was, it was fun to talk about Sun and [01:13:15] then like that, uh, the kind of like how thing got set up, uh, at, with, with the Kubernetes community. So that was good [01:13:20] stuff. Well, it was, that was great. And, uh, yeah, now people know where to, uh, where to check you out. So, [01:13:25] uh, hello listener. You've been listening to Software Defined Interviews and, you know, [01:13:30] Josh just listed out a bunch of those things, but if you wanna check 'em out, you can go to software defined [01:13:35] interviews.com/ 1 2 1 and find it there. [01:13:37] Coté: And, uh, with that, we'll see everyone next [01:13:40] time. [01:13:41] Whitney Lee: Thank you so much. Bye. [01:13:43] Josh Berkus: Bye.