Greg Sadetsky, Antoine Leclair - Disco

{/ TAB: SHOW NOTES /}

This week we talk to Greg Sadetsky and Antoine Leclair, the creators of Disco. Disco make running you own infra a piece of cake.

• https://disco.cloud/
• https://docs.letsdisco.dev/
• https://github.com/letsdiscodev/
• https://github.com/gregsadetsky
• https://github.com/antoineleclair

Episode sponsored By WorkOS (https://workos.com) and Mailtrap (https://l.rw.rw/devtools_2)

Become a paid subscriber our patreon, spotify, or apple podcasts for the full episode.

• https://www.patreon.com/devtoolsfm
• https://podcasters.spotify.com/pod/show/devtoolsfm/subscribe
• https://podcasts.apple.com/us/podcast/devtools-fm/id1566647758
• https://www.youtube.com/@devtoolsfm/membership

{/ LINKS /}

{/ Paste show notes /}

{/ TAB: SECTIONS /}

[00:00:00] Introduction [00:05:22] Challenges and Rewards of Freelancing [00:12:19] Ad [00:13:05] Introduction to Disco [00:14:51] The Origin Story of Disco [00:24:45] Positioning Disco in the Market [00:37:10] Convenience vs. Cost [00:42:59] Raspberry Pi Hosting [00:48:09] Open Source and Business Models [00:57:30] Conclusion and Future Vision

{/ TAB: TRANSCRIPT /}

Greg and Antoine: I think it is interesting to see those cases of the people losing the lottery. In that they deploy a project and the thing goes viral, and then they get a bill and I'm like, oh, you were popular, so now you're in debt. That sucks.

[00:00:19] Introduction

Andrew: Hello, welcome to Deb Tools fm. This is a podcast about developer tools and the people who make 'em. I'm Andrew, and this is my co-host Justin.

Justin: Hey everyone, uh, we're really excited to have, uh, Greg Sadetsky and Antonie Leclair on with us today. So, uh, I know Greg from Recurs, uh, Greg, you're very, very prolific at Recurs. I've worked on a lot of cool projects. I'm pretty excited to talk about some of that. Um, but also you're working on this project called Disco, uh, and we're excited to learn more about that today.

Uh, but before we dive into that, would y'all mind, uh, introducing yourself to our audience and telling us a little bit more about yourselves?

Greg and Antoine: Yeah, for sure. Extremely happy to be here. Hi everybody. Uh, Antoine, please go ahead. Thank you. So yeah, I'm, uh, Antoine and you send my name, my name, uh, correctly, I guess Antoine. Um, so yeah, I've been, uh, doing software for a while now. Uh, before that, like, uh, early two, 2000, I was a aircraft pilot, uh, but switched career to, uh, being a photographer, like a press photographer for a few years full time.

And then, uh, I switched to, uh, software development. Uh, so I've been doing that since like. 2008. Um, I worked at a few smaller companies, uh, building software for other companies like, um, uh, consulting. Um, and I've worked on a few SaaS products as well in the past, like in the years, like, uh, 2010, 2013, around that time, uh, I worked on the one that, uh, we sold like in the 2014 for, uh, e-signature that, uh, uh, failed against, uh, like the bigger names in the.

Industry, um, also started like a, um, uh, thing like, um. yeah, it went great. Uh, product, I, I left the product though, but it's been sold recently by the guys that left the, uh, that was that there. And, uh, successful product. Uh, then I spoke to freelancing. I went with, uh, 10 X management, uh, where Greg, uh, was also for a few years.

And, uh, I've been with them for, uh, uh, big wine house since, uh, like 2013 I think. Mm-hmm. Uh, I've been on many projects with them. Uh, super nice place to do financing and then put you in touch with, uh, great clients. Um. And yeah, since maybe 2016, uh, I've been working with, uh, also through Tenex, uh, with a great client that's called, uh, ideas.org.

It's, uh, the biggest, uh, nonprofit job website, uh, on the web in the us I think. Um. And, uh, it's a quite, uh, you know, large traffic site, uh, like the perfect size for a website to work on. Basically, like not too big, not too small. Um, goldilock website and, uh, uh, right now they are merging with, uh, volunteer match.org.

They're kind of swallowing them. Uh, it's the biggest volunteering website in the US also. So like a pretty big merge. Um. yeah, for them, uh, I've worked on many, uh, big things, many small things and uh, yeah, here I am. Yeah. Um, yeah, so I'll complete it a little bit 'cause yeah, our story definitely inter interwind a little bit. Um, yeah, I've been doing software for a while. Um, I also for some reason have a lot of friends who are pilots. I think, uh, they compliment well. Um, they're very calm and, uh, it's great. I, I have like a really great friend who's in helicopter pilot, just a great crew. Just, you should make more helicopter. Uh, just more pilot friends. I recommended. Um, but uh, yeah, so I did software for a while and yeah, did like a startup some years ago here in Quebec where actually where I am right now, where, uh, Awan lives and my parents live. And, uh, yeah, I met a bunch of people, had like a coworking space for a while. I. Uh, did like a master's degree where I did sort of computer vision and did like biology stuff.

It was super, super fun to see kind of all that world. Um, and, uh, and then, yeah, I was also, I've been doing freelancing kind of for a while, um, through 10 x, which is 10 x is the, like we really get to say a really cool thing, al 20 I, that we have an agent and the story is that. The 10 X people came from the music world where, and I think it's totally public information, but like Yeah, like one of the agents represents like Vanessa Carlton. So it's like they're like agents, agents. But they also saw that in interestingly, developers can have this sort of same, I. of personality then, like musicians where they kind of really care about the craft and they don't really are like that interested about the contracts. They're like, yeah, yeah, sure, sure, sure.

You know, but it turns out that when you say, sure, sure, sure. And somebody's like, oh, great. Like, I'll make you a really bad offer. But you'll say, sure, sure, sure. Uh, you kind of can get, not necessarily taken advantage of, but it's just sort of like, you know, you not kind of like be as, I don't know, like the contracts might not be like as, as well written or whatever.

So basically they're an agent, right? They do that and the, uh, they, they do that for. Uh, software developers. There's a, a bunch of people on the roster. Um, I don't think we met like that. I think we started working, uh, on a, on a bunch of projects together. Um, and I definitely also worked on the Idealist Project.

Um, yeah, just like really beautiful, you know, fantastic, like, as the name says it, um, uh, you know, just like a great side, great people. Um, and then yeah, but also a lot of other things like recur. so, uh, which I'm happy to talk about, uh, as well. Um, and yeah, I was just as a startup, like doing some mapping and, uh. But yeah, touching lots of things is a bit of my story.

Justin: Y Yeah. Uh, it's, it's kind of, it's cool to hear your combined stories.

[00:05:22] Challenges and Rewards of Freelancing

Justin: Um, uh, I, the freelance world is such a different world. Uh, you know, it's like generally just keeping on top of everything. I think it's, it's better if you work with like an agent or an agency or something where you, you sort of like.

You get to think about the, the craft or like turning, turning the products, but the, the freelance world in general seems like really challenging. Uh, how's your experiences been with that? Like, uh, I mean, it sounds like you're sort of working with a group, so it's maybe like a little bit less admin overhead, but like what is the, what

is that world like now?

Greg and Antoine: Yeah. I think it's kind of like not only is it not for everyone. Like some days I'm also like, is it for me? Like, it's just like, it's, it's quite different. I, I actually, uh, just had uh, a very like way more kind of like a nine to five job, like at a, at an incredibly great place. Uh, you know, had colleagues at everything and it's sort of like, I was like, oh wow.

Like I, I'm getting to live the whiplash. 'cause it's like this is the opposite of what we do. And so of course there's just moments when you have a job where it's like. Oh, so many things are like not part of your daily kind of thing. It's like you, you know, you mostly think you're co your, most of your colleagues are gonna be there the next day.

Uh, there's like a lot of things that you stop sort of questioning or being in this mode when you're a freelancer. I think you're just way more of, you know, you, you also are representing yourself. Uh, you are kind of like. You do have a brand of you, you are selling yourself. Uh, you're actually going through like a micro interview with every customer. Um, even though like the agency does a good job of like selling us and kind like talking about how we're like incredible and everybody should work with us, but still, you get on a call with somebody who's never worked with you and they're gonna give you GitHub access. so they kind of, you know, they ask you a few questions and so it's interesting 'cause it's not like a lead code, so you, you don't get that pleasure of having to solve like puzzles.

Uh, that immense pleasure, that just thing that I miss every day, um, of my life. But, uh, but you do kind of get, you know, a little bit sort of like, yeah, just get kind of a few questions. And so it's just interesting 'cause then you look back in a year, if you've talked to like five or 10 clients you've interviewed 10 times in a year and we've been doing it for 10 years.

So it's like. Like, you know, at some point I was just like, yeah, I guess I just like interviewing now. Like this is just my new thing. Of course, when you have a job, you, a lot of people hate that and they just want to keep their job and do the work and not think about interviewing or selling themselves or having an online brand. So, um, the, but you know, um, Antoine can be like a freelancer, but also stay with a client for a long time. My kind of freelancing was also a bit more like maybe some contracts were smaller or, or larger or sometimes it really, somebody had like an algorithm problem. I still remember a client who like just wanted us to build a better sorting algorithm and I was always like, are you sure?

Like you wanna hire somebody, like sort what was, do you remember? Oh yeah. 'cause it, it was an app where they wanted, like, they really wanted their users to be able to rank the things in an order of preference, but they were like, but there's too many items. So I don't, I can't have them like actually sort, so they should be presented with like two options and they'll say which one they prefer. And I was like, that's good. But then if we looked at the numbers, they would have to do that like 1000 million times. Like it still kind of worked but didn't. And they were like, no, that's, and I was like, all right. And I was like, it was kind of exciting to build this. But then once we got like, it was literally like we, you know, it's like I emailed the JavaScript file and I'm like, goodbye. And they were, and they were gone. And I was like, that was weird. But, um, you know, that, that also can be freelancing, um, which is fun. And, and, and for sure the sort of like. Uh, more like wanting to see other systems and kind of touch a lot of things. And, and it, you know, it's this feeling, what is it in New York?

The open doors, you know, the, like one day in a year where like you're allowed to go into like the sewers and see like the plan or whatever. It's like one day, like you get an Eventbrite and all the tickets sell out. It's terrible because ev I, I would love to do every item of, of this, and they all just sell out in a second.

But like for one day you get to see the secrets. And I think freelancing has that aspect. Like, I worked at Verizon on a streaming platform. And for one second I was like deep in the Chromecast analytics code and it was great. I was like, like I love Chromecast because of that. And I know a ton about like M three U and you know, when somebody like, is like, what do you think ofms.com?

I'm like, they're extremely cool. Like I've, I've, I've touched their thing. I was like, they, they can be really trusted with everything. Um, and that's cool. And then another day you, I, you know, we worked with Wall Street people and they were doing stuff that I had never seen. So yeah, you kind of get a little bit of that sort of like, ah, you kind of open a door and see a thing. But after the contract's over, that's it. The checks stop and the, the insure e everything kind of stops and you're on your own again. So, uh, yeah. Also I think it, the, the way you see work is a bit different, like compared to an employee, like often. Uh, when I see more work for me, it, it's a cool thing 'cause it means that I have more work to do and like, you know, I, I get to be paid for longer.

So like, uh, I, I see big projects, complex projects as a, as a nice thing, and I'm super excited to attack those projects. Mm-hmm. Every time, uh, uh, there's a new thing on the table, I'm like, yes, let's, uh, let's do that. the employee just sitting next to me is like, uh, well, well, not where I work, but like, I've seen that in the past, like in other organizations where like, eh, you know, like, uh, will we be done, like building things.

I would just like to, to uh, to relax a bit, you know, like, but when you're a freelance freelancer, you can, uh, that is interesting. You're excited, uh, at those products, I guess. Yeah. That, that, that is a good point of view. Yeah. Yeah.

Justin: Very different incentive structures.

Greg and Antoine: Yeah, yeah, yeah, yeah, yeah. But, but they really both like, they both make so much sense.

Maybe because they're so different. Like

Justin: Mm.

Greg and Antoine: we're employees or like when I was an employee, I was like, oh, we should absolutely have this person come in for a couple of hours. It makes so much sense to have this ML expert come in. Uh, and I know we can't hire them. That's fine. I'm not asking for budget.

I'm just like. I know you have $300, like let's just hire this person. Um, but I also could see that I, you know, at one time I was like literally giddy with this incredible who was like, less than half my age, who had come from like Waterloo and we had just like hacked the hex bites of like some thing we had reverse engineer.

And we were like, we were like shaking and I had this moment, I had like calm, Greg was like this, like when you're freelancing remotely, you don't get giddy. Like you don't hold hands with your colleagues and your giddy. It's like this is a very, being at work office, you know the other person, you kind of trust them.

You have this like rhythm and then one day there's like this moment where you're looking at hex bites. And that's, that's a little bit of a, like, I kind of have to have a job to have those moments maybe, you know? Yeah. So it is just very different. It's cool to experience both. Also the way I see it, like, you know, like, uh, in the cloud, how you can scale up machines by sliding something.

I see freelancers like that where it's easy to scale up and down, and like I. The way I see it, you have to, to give a good reason for the, not the employer, but like the, the organization to, uh, not scale you out, basically like to, to keep you so like they, they have to get good value. So it keeps you like, uh, up to date and like always proving yourself.

And it's, uh, for some people it could be tiring, but, uh, I feel like it's, uh, also exciting to always be like, uh, thriving and trying to, to build cool things and, uh, doing quickly and in a cool way. So, yeah. Yeah.

[00:12:19] Ad

Justin: This episode is brought to you by WorkOS. WorkOS adds enterprise features to your app without the overhead.

Single sign on with any provider. Directory sync that just works.

Role-based access control, audit logs, admin portals, secure credential store, you name it. They have modern APIs, well-designed, SDKs and documentation that respects your time. It's built for developers, backed by great support and trusted by teams shipping fast and scaling up. So just focus on your product and let WorkOS handle the enterprise stack.

You can get started at workos.com.

WorkOS also has a podcast called Crossing the Enterprise Chasm. It's really worth a listen if you're in a startup and you're looking at serving upmarket or moving to enterprise customers.

Thanks again to the WorkOS team for sponsoring us.

Andrew: Uh, s

[00:13:05] Introduction to Disco

Andrew: peaking of building cool things, uh, you guys are working on a pretty cool thing called disco, so could you tell us what disco is and what led you to creating it?

Greg and Antoine: Yeah, for sure. Um, I'll let you maybe wan mention some of the things that led you creating Cool. The, the, yeah, I think the. Yeah. What is disco? Okay, well I'll start, start there. 'cause it would be really weird if I started the history of a thing that nobody knows. Okay. So disco is a, is a platform, is a, is a bunch of quote code, uh, that's open source that you can use to deploy and host web apps or projects. So, very roughly speaking, there comes a day and every developer's life or they're like, ah. You know, it could be cool to actually share my thing and put it in a place and not have to touch it. And other people can come and see it and have a URL. That's roughly what kinda disco just simplifies. Um, there's a lot of steps, sometimes way too many steps in getting that to work.

Uh, and to like, you know, like stay up. Um, there's a lot of kind of like check boxes. It really feels like, um, I, I make a ton of analogies, which I shouldn't be allowed to aviation wise 'cause I. Truly, you know, am not like a pilot, but to me it feels like a checkbox. It's like a list of checkbox. It's like, oh, as my whatever thing off. Um, thanks for helping, uh, flares are down. I don't know, are the gear, like whatever. You kind of go through a checklist. But the thing is, for every web project there's a checklist. Oh, did you forget that this, is there a, is there a thing? How are you thinking about a future thing that will inevitably happen but you don't wanna be thinking about it?

'cause now it just runs a local host and it's really fun. And so once you go past deploying 30 or 40 things, it's extremely not fun to redo the checklist and to do manual things. So at, at some point, of course, like any developer, you start automating it. And so disco automates most of the things that you kind of just are boring or uninteresting or, you know, make your side kind of work. So that's that. Um, what led us there, I think. Yeah.

[00:14:51] The Origin Story of Disco

Greg and Antoine: So I gave a ization Recurs and I kind of like thought. Deeply, kind of like what did lead me here? Um, 'cause it, it is, it is like we, wanna say weirdly a project that I care about. I care about every project, but it, there is, there is like a, a weirder, even like more caring kind of thing.

I'm like, wait, why, why, why does it matter to me so much? And it turns out I looked back and I was like. I've been doing websites things since I was like a teenager and I literally had a server in my room growing up that I know, like inflicted, like terrible ire, irreversible damage on my sleep schedule.

Just like I always have a fan in the background, but I had my own server. I remembered the IP of that server. I'm not saying it out loud, but I, I a hundred percent remember the IP of that server. Uh, which is also kinda a weird thing having a static IP when you're young anyway, but I was very lucky with that. Right. And, and so, um, yeah, we've been doing web things for a really long time. The story of how we started disco. There's a lot of like smart things I will say. But the funny thing is that Antonin and I so wanted to kind of do a thing together that we had this Google Docs, which I'm very proud of. I don't know if you're ashamed of it.

I love it. It's 50 pages, five zero pages of Antoine and me being like, what about. And it's like the most random, like nothing is more random than that document. Like at some point, like, like I know there's a lot of like weightlifting stuff. 'cause obviously, you know, one of us like actually cares about their body. Um, um, there's like a ton of like we could track and then at some point like AI comes in and it leaves and it reappears again. We could have logging things. I even kind of registered maybe a domain that we abandoned. But it was 50 pages. So I know there's a lot, like there's a conversation we're both kind of gently telling the other, I'm not sure about this 'cause I've never like done. And so we kinda back up. But at some point, and of course the joke is it was on the first page, which Twan wrote the, on the first page, Twan kind of goes well. I have been deploying a lot of things and writing a lot of glue code. I think that it could be a little bit automated and simplified. So that's kind of how I see how you came to it.

But how did you actually come to it? Yeah, so I can explain like the way, the way I had this idea like, uh, in first place. So, uh, at my, uh, with my client@ideas.org I mentioned earlier, um, uh, since like 2017, they, they've been hosted on, uh, Heroku. Uh, it's a nice platform. Uh, it's working well. Um, and it. uh, it got a lot of love in the past, uh, because they solved the problem back in the days, I don't remember the year, but like maybe 2012, something like that when they started hosting like a Ruby on Res app.

And yeah, they simplified the process a lot by a lot, uh, because it was very complex back then to deploy an application. Uh, and they came up with the 12 factor, uh, apps and. Made everything like a mo, like a move the humanity one step farther, uh mm-hmm. In the deployment world. Um, and, uh, IDS was, uh, was set there.

And, I would say though they, they've been bought by, uh, Salesforce and they kind of stopped, uh, innovating at some point. Um. We were still using them. They, they were still, uh, working great up to a point when, in maybe 20 20, 20 21, I forgot exactly when, but like they started to have a, a bunch of. Problems like, uh, downtime, uh, the support was not good even, even though we had like enterprise, uh, contract with them. And, uh, I, I always had like to escalate the, the same question like five times before I actually talked to someone that knew, uh, what I was talking about. Um, so, uh, I just started to look into like alternatives, uh, to somewhere else basically.

Um. So, uh, back then there were a few alternatives starting to show up, like render was in the early days. Um, and, uh, otherwise Kubernetes was around, uh, Docker Swarm was around, um, and Doku like the open source, uh, platform to deploy on your machine. Um, other names like Cap Rover, things like that. Uh, so I tried a bunch of those.

Um, and it was never like, you know, the, the, the perfect thing for I Ideas, like right now, I did, uh, for that client, they, I. They have like a, maybe like se seven different services running in the, the main production project, and they have like a, a few different projects running on a Roku and. E either like things were too simple like for a team to work on.

Like, uh, for example, doc who, uh, if you use that, you are SSHing in a, in a machine, uh, you install stuff yourself and if you want to manage a team, it's not really possible because like you won't start to like, uh, add and remove SSH keys on, on the machine and things like that. Uh, and also it was hard. Or not possible to scale.

To scale. Yeah. Horizontally. Like let's say you have a machine and it's not big enough, like you cannot add more machines to it. Um, and then, uh, Kubernetes, I tried that, but like, you know, like the, the, the size of the team that didn't make sense to use Kubernetes. Uh, the team is like, uh, six developers.

Uh. Approximately. And you know, like, uh, Kubernetes, everyone I ask like, oh yeah, you're using Kubernetes. They, they all said like, oh yeah, it's, uh, it's super simple to use. You should try that. Like the the hosted ones, like on the digital show, the AWS whatever, like, uh, and I tried that and. It felt super simple at first, and as soon as I started to look into the details, like how to set environmental variables, how to like do this thing securely, how to network, things like that, then the complexity was huge and it was almost, uh, impossible to manage for a small team. And I feel like most small team using Kubernetes today, they just ignore about, uh, about all of that. And they just like, uh, have something maybe that's not secure or something like that. And, uh, about that time also like. The, I think it was a Department of Defense in the US that came out with a huge, like, PDF of like, uh, maybe 20 or, uh, I don't know the relevant number of pages, but like a bunch of pages of things to look into to secure, uh, Kubernetes deployment if you wanna make sure that it's, uh, it's secure. So even though like a. At first, at first sight. It, it looks like it's a simple thing that will scale up and, uh, and it's great to manage and it's, uh, it's like getting started is, is easy. But as soon as you look into the, the more complex things, it's, uh, pretty complex to, to deal with. Uh, so far a small team, it didn't make sense.

Uh, I. And yeah, so basically n nothing made, uh, really made sense and it felt like, yeah, maybe then like I, I should just write like some glucose that would like, uh, you know, use Docker or something in the background. Maybe Docker, swarm and, uh. To deal with that specific project. But then like, uh, I decided to not move that forward because it would, it would have taken too much time for that one client and uh, I decided to just like leave them on a Roku basically back then, because it made the most, the most sense for them, for that small team to, uh, to remain there. But still, I had that idea in the back of my head like that. Yeah, maybe I should write like some project, but that is that. Blue code basically, that every organization has the right for to deploy their project and like not have to like do it all over again for each organization. Um. yeah, that was one of the idea and the, and the huge list of, uh, of ideas with Greg.

And at some point I decided, yeah, I'm, I'm paying like a, a few bucks a month for, uh, my own like, personal websites that are just like static HTMS sites. Nothing complex. I just thought like, uh, yeah, it's the holidays. Like, let's take some time and put that on the raspberry raspberry pie at home. Um. I tried like caddy the web server and I was like, oh yeah, this is cool.

It's, it's much easier than like back in the days with Nginx having to buy SSR certificate and all that. Mm-hmm. Mm-hmm. You just like set your domain name and you have a TS certificate set up automatically. And it was pretty easy to get this, uh, set up, but then like it didn't, uh, do all the like, uh, git uh, deployments automatically with GI push, things like that. So then I started to look into like. idea, I checked with Greg again. Like, would, would that make sense to move that idea forward because it seems like a cool idea. And then we started work on that, basically the, the disco. So yeah. Yeah, yeah. And, and also the context of, uh, yeah, Justin, when you were asking about recurs, uh, which yeah, folks, everyone should, should check it out.

It's an amazing place where people go and spend usually three months and working on kind of personal projects. When I went in. Um, I had this very, like, it's very scary 'cause a three months go by in truly like an instant. And I was really scared. I was gonna spend a lot of meta time, like not doing stuff, but like worrying about my tooling kind of. And so I kind of gave myself like a rule, like, okay, I'm gonna take three days. I'm gonna figure out how I'm gonna deploy every project I'm gonna make. Because I, I did kind of make a lot, just a lot of like, just like simple, small things. I would deploy really, really fast. Sometimes it's like a little, you know, it's kind of like, almost like it's more, usually the concept was more interesting or funnier than like the tech.

It was just, I, I just need a database. I just need a Python file, but I also need a place to host them. And so back then, because this was a predis kind of time, a couple days also like looked at Doku, like kind of things worked or didn't work. Uh uh. I was kind of unsure and it kind of made me nervous to bet on something and so I took render, which was the big Heroku competitor and still is I. Um, and I, I just like standardized my stack and so I kind of made a file that I almost like just duplicated every time that I, I kind of like deployed like about 20 projects. Uh, and it was all just like, I'm just doing render. I'm just doing render. Uh, and then I, Antoine said, few bucks. In my case it was like 300 bucks every month. And my list of projects are idiotic, like one of them is called Restaurants in Peace, restaurants at RIP where you can leave reviews for restaurants that died. That's stupid. That's funny. That's like a joke. That's like a chuckle, right? That's just something you just like, but is that worth $29 till the rest of, you know, per month for, for eternity, you know?

Um, and then I have like 10 of those and um, and so back then I was like, well, yeah. Like, yeah, for yes. You know, Antoine was like, oh yeah, doctor this Will, I'm like, yeah, you, are you saying I could host everything on one server? That sounds lovely. And so, uh, we had this kind of like real interest, and of course this.

Completely goes back to the server that I had in my house when I was a teenager, where I had Apache. I did not have s. SL, didn't get push, you know, just FTP, good old days like we used to do. And I just had one server, obviously with my static ip. So yeah, so that was kind of the origin.

Justin: Okay. A

lot to unpack there.

[00:24:45] Positioning Disco in the Market

Justin: Um, so the hosting space has changed a lot. Uh, I think there's this like classic trope of a lot of like, companies building open source tooling and you ask them like how they monetize and they just say hosting, like, like they have some like hosted compute product, that'll be their like thing, and they'll, they'll figure that out at some point in the future.

Uh, but we have seen like. A little bit of a renaissance of hosting platforms. So you mentioned render, uh, which has been, uh, a big one in the space. Uh, there's railway, uh, which we've had on the podcast in the past. Um, so there are, there are these like. Heroku alternatives. And then we also have the, the sort of like serverless world, you know, started with like AWS but like it's more, I think often of like CloudFlare or ELL of like very much what is the minimal amount of infrastructure that you need to think about?

How do you just think about the code that you're deploying?

Mm-hmm.

there are also like ecosystem constraints to those a lot. So it's like usually JavaScript, for instance. Um. So if you're, you know, more traditional, like full, uh, backend framework or whatever, there, you kind of like are pushed out of those environments.

Greg and Antoine: Mm-hmm.

Justin: Uh, and I guess I, I didn't mention FLY io, which is another, uh, important one in the space, in the like, kind of Heroku space. Um, so. Uh, you're, you're talking about kind of like disco being, uh, good for like personal projects, small projects, and it seems like the more like Heroku replacement, uh, so like how do you position it in the market?

Like what, like how do you say, okay, disco is a good fit. If you were thinking about like using fly.io for example, and instead of that you could use this in these cases, like how do you, how do you sort of like pitch it to

people?

Greg and Antoine: Yeah, it's a good question. think, um, know. figuring it out. Uh, you see the team here, uh, we've been both like working on it, but also thinking about it. But, you know, um, um, um, I'm not sure that there's a answer. I think I don't have, I don't, I don't think that there's a universal answer, which is also probably not what you're asking, but I think the one thing that's hard is to say, well, obviously every single web hosting project should be on disco.

Like, that's just so clear to me. That's, that's where like, good to go through this exercise, which is like, no, we don't think that. Um, we actually are kind of like more also into the idea that, um, you know, there's a ton of options. I mean, you had the HH on the PO on the podcast, right? And like, you know, they've got Kamal, right, which is this great option.

Um, that's a little bit more like Ruby and kind of like rails oriented obviously. Um, but quite close. Like there's a lot of like similarities. I was looking at their command line, I was like, oh, this is really neat. We have kind of similar things. Uh, they also went like with Docker files and things like that, so I think. know, this is more about us being the weird new flower in the renaissance, the garden renaissance of like this kind of like all world. Um, we are looking for like, what is the exact sort of like, oh. 'cause it's like, it's an eight dimensional thing, right? It's like you'd have to know where you are on eight dimensions to be like, I think disco would be kind of perfect for this.

So it is hard also for us to like say it in the way or just like right now we don't have that perfect color period pitch, which is if this, but not that the. Simplest or the, you know, in that eight D blob, like the one blob that I can sort of like see the most is at Heroku you're kind of paying so much that the ratio of what you're getting to, what you're paying to what you're using, that's not necessarily out of whack enough, but you're like, oh, like that could be replaced.

And then, I mean, because we've had this, the experience of moving like. Big staging and we can talk about that as well. Uh, big staging things off of Heroku. Um, we kind of fit perfectly there because it's like, it's pretty much one to one. Um, you kind of only get advantages. The infra is still at AWS so it's like reliable enough, the cost go way down. Um, that works really well. We're also interested in that a, like, we're trying to make it work as a, you know. Is it a startup? I guess it's, uh, of course, yeah. But like the, the startup shape of what we're trying to do, the managed, like, we're trying to also like sell a service. We're also interested in people that are like spending a lot because it's just like, we kind of feel this is also like how I go sometimes with personal things or asks or projects.

Like, you know, I feel like if a friend asks for a favor, I mean a friend, you never charge them. But just like I feel it always easier to be like. Yeah, I do free things for people who are like in my neighborhood and my like, and then when, like the giantess most biggest company comes around, it's just, it's okay to charge them 'cause they have money.

You know, they're making money. We're making money. They don't make money. I don't charge them. And so to me it's this, again, I'm kind of applying, I think we're applying a bit of the same model. It's like we're open source and we're open for people who are Yeah, both hobbyists or, you know, Mike, you know, the, the hobbyist pro, whatever, like lots of sites.

Uh, there's also somebody from ERs. Uh, um, uh, I don't know if they'd love to be called out, but they're, they're great. Um, their first letter, no, that's even too much. Whatever. Their first letter starts in the second half of the alphabet as well, I'll say. Uh, and so try to guess. Uh, but they're really amazing and I know that they're running 70 projects on the box and we just had this call where we're like, oh my God.

Like we were just so happy to be like. The, it, it's so funny, they were kinda looking at us, was like, yeah, I, I, I'm aware, like I'm aware of what I'm doing versus what I could be doing with this. So I was like, okay, that's interesting. That's like a clear case, uh, because again, they're just paying a lot, uh, in management and, and everything. But on the commercial side, I think it's, yeah, it's kind of like fits us Is this Yeah, this sort of Goldilocks that may, again, we might have to kind of define in a better way because it's a little bit of like. You know, I don't know if people should try us first and then feel it out, whether it it fits their case.

Uh, we're even talking like, you know, why would you, what would it be a better idea to go through Kubernetes? And it's something like maybe if you have like more than a hundred microservice, like there, there, there are, I don't know if there's like numbers there. Maybe it would be good if we had numbers.

I don't know. But it's like there's like a scale of a problem you're solving where like you'll have enough really complicated networking between your services or like Yeah, Anton was reminding me like, on Heroku services can't. to each other. Right? It's like apps, like different apps, right? Different that you deploy can't, like, they don't have like an internal network and obviously other companies like maybe I think Fly has like more of a notion of this network.

We are way more of a, like if you deploy it, it's all kind of like they all kind of talk on this private thing through the Docker Swarm. So. Things like that. I think where it's just like where there's like a better fit, uh, is a little bit easier. was an interesting case. I'm helping a friend right now, recurs. Oh, I think I can name, I'm starting to be so cagey. I just never know if people are like, oh my God, you named me. Um, I feel just bad. But, um, also it's great to name drop. Um, it, so, uh, yeah, my friend River who's working on an instrument and he's working on an instrument in C the instrument is like just incredible. he's recompiling in Wasm and then we're doing like JavaScript stuff on top of that. And the joke was, he was like, well, I'll obviously just deploy this to vercel. It's, it's, it's obvious like everybody does vercel. And I was like very supportive. I wasn't pushing, I was like, yeah, great. Like I, I wasn't like, you have to use this code. it was like, well wait a second. Like, can you run a c wasm m script and compiler as part of your build pipeline on versa? like somebody might be like, yes you can, you person. Um, and so you're great. Uh, but if no one says that, then I am not that stupid. And actually, I don't think you can, I.

And it was like, we're trying to shove this like, you know, like this build script. And then at some point I felt really bad 'cause I'm not probably good at whatever. I was like, do you mind if we, if I just write three lines in one docker file, I promise. And I am not good at docker files. So it was like, I was really like, just am scripting a PT get. And then when it worked, I really did have this moment of like, this is that kind of Goldilocks moment that maybe, again, we should try to like, you know, really explain well that it's just like your project is like. Complex enough that it's kind of real. Think GitHub pages still makes sense if you're only doing a like index. I think our side. Is it? No, not, we moved over to really dog food, but for a while we had like a GitHub pages for our official website. 'cause it's fine. Like if you're really, truly like, just serving static, completely static files, it's a great option. Um, yeah, the way I see it a bit, like I said earlier, like a, I see it like a gradient, like a, one side you have like a Kubernetes or like totally custom deployments and like terraform, things like that.

Yeah. Uh. For very complex projects, for like, uh, big teams. Like, uh, like you said, like maybe like you have a hundred or maybe 800, uh, microservices that need to be deployed separately by separate teams and they need to talk together. And that's a totally different, uh, requirement that like disco doesn't try to tackle. Um. And on the other side of that, like I feel, the other extreme would be platforms like vercel for example, where you don't really, uh, see the infrastructure. It's not something that exists. Um, and uh, on Heroku, uh, you could deploy there, but then maybe you lack some control, like, for example, services talking together.

If you want like a, if you have like your web service, for example, talking with another worker to. Process something. For example, you'd need like a message queue to communicate together like that. Or if you want to do like a, a server side rendering, for example, you couldn't have like two different services to do.

That would be, it would be kind of counter complex with this code. Pretty easy to do. Mm-hmm. Uh, so that, that's one thing. So I think that that's kind of a, the spot where we, we are is kind of, uh, just like a. Between, uh, Kubernetes and Heroku? I would say more closer to, uh, Heroku than Kubernetes. And also I think there's a, a big cost saving in a sense, uh, compared to a platform like Heroku. Uh, because on Heroku, if you have like a, let's say a project you have, uh, they call them diagnosed, but let's call them like a servers for example, or services. Uh, if you have like a, a few services running scaled, uh. Two, uh, like two instances, for example, then you might, you might have like, uh, 20 services running, uh, you page for each one of them.

Uh, a, a large amount. Um, but for, uh, scope, for example, you could all run all of them on a single machine, uh, for way less than that because you don't pay like the overhead of like, uh, paying for that extra padding Ram for each of the instances. Uh, so that's one thing. And, um. Also the fact that like you, you can deploy.

Yeah, exactly. Deploy many things on, on, on the same machine basically. I think you can, you can say yeah, that. Uh, and I guess I can talk about like the staging environments and things like that. Uh, so with, uh, my client, uh, ideas, uh, what we've done, uh, recently is that, uh. We moved, uh, we started with a staging environment to a disco, uh, on Heroku, having the staging environment running with, uh, all the dynos database and things like that.

It costs around like $500 a month. Um, per staging. Per staging environment. Yeah, exactly. So we moved that to like a, a single machine that cost like a. 50 euro is like ner for like a 50 year a month. So like already it's like a 10 10 x sa saving, uh, for, uh, one staging environment. But then like, because uh, the machine was still like, so unused, uh, I was able to add like, uh, I think seven other staging environments.

So at that point, it's like on the Roku we wouldn't have done that because it would've cost way too much. But now, now it's, uh, it's possible to do like, it opens new doors because, uh, it's not scaling up the pricing because you are using more, uh, because you can use that. Padding basically. So like the, the seven, seven staging environments would have cost maybe $3,000 on ku, but now it's still like 50 euros for all of that. And, uh, recently I had more projects like, uh, just like, uh, you know, profile concepts and things like that just to, to. Not prove a point, but like sh showing something to someone else, uh, I just created a project, deployed it to this code was super easy. No, the, the question of the pricing never, never came to mind because the machine was already, already there, already paid for 50 rows a month, and I could just add a projects in like, uh, 20 seconds and the project was deployed and you get pushed.

It deploys the, the updates. So, uh, it's very easy to work with. Yeah. But I, I think, yeah, the, the sort of, the question of the, the platforms I, I. I, I, you know, I don't know how kind of fundamental the business model is to the business. I mean, it's sort of absurd to say that, but it's kind of like, you know, it really feels like Amazon leases you infra and they're like, here's bits.

You combine them, there's atoms, more atoms. You pay more for them. If they're bigger, you pay more. But they're also depending on the product, of course, that they have. But a lot of them, they sort of, they get out of your business. They, they're not really like, you have to assemble those things, but they feel more like Legos.

[00:37:10] Convenience vs. Cost

Greg and Antoine: And I think, yeah, the, the real, real thing with everybody else a little bit is not that they have like a suspect business model, but it's a bit of like the convenience got so convenient that they started selling the convenience. Like the convenience was the kind of the, the. The candy and it's like, you know, like they kind of, you know, this is the thing.

It's like, I'm like, I understand why and it made sense. Like was the thing that made Heroku Heroku, um. it's kind of like, because it deployed fast, it's sort of the, the, yeah, the flip side, you were like, oh yeah, you like that. Like it was real fast. Right. Well like, well, yeah, you did it once.

They're like, wait, you're counting, you're counting how many times I'm doing that. And obviously, you know, like I know they don't charge or deploy, which maybe I'm, I'm hope I'm not giving anyone ideas, but, um, I think it is interesting to see those cases of like the people losing the lottery. In that they deploy a project and the thing goes viral, and then they get a bill and I'm like, oh, you were popular, so now you're in debt. That sucks. That's of the opposite. You, you should have become popular now you're poor. It's like what? Um. yeah, because, because at some point you, right, right. Like the, the, those limits are more, it, it's kind, it's kind of stranger to, I, I feel read a pricing page that has, this is a personal thing. I'm not saying anything. This is why we did our thing, because we find that weird. Somebody else might not find that weird, but like. There are so many of these moments where like, ah, you, you're doing it, you're doing how many edge function in vocations. Aha. And it's like, then there's a number and you're like, well, did I go over 25,000 or a hundred?

Or, you know, um, I remember a Nolan royalty who you, you might know who of course famously, you know, did 1 million check boxes and, and uh, you know, now he just released one 1 million chess boards.com, which is incredible. he definitely, you know. Has a thing where he kind of keeps deploying these things that have like the most extreme scale.

Like it's, it's just a barrage of users and coming and actions and he keeps kind of like scaling down his infra. I mean, he's like on a mission to prove something. I think at, at, at, at the end it will just like run on a stick. Like he just like take a stick from a branch and he'll somehow like connect it to wifi.

No, I'm watching you. But, um, but he for sure, I've absolutely seen him think and talk and I'm, I'm even part of a talk that he gives because. At some point he tells me that this thing is going really well, and I go, how's your bandwidth, buddy? You're at Digital Ocean. There's a number that you're not allowed to hit. is a number of bites, uh, which we also have bandwidth limits. I'm, I'm not saying this is some sort of like, we have a magical deal with God, where like bandwidth doesn't count, but you know, he was just like, that was one thing where he was not allowed, like, it's kind of like the nuclear reactor, like you're kind of turning the knobs.

There is a red zone where you suddenly start getting billed. And of course other platforms are just like they've, they've. Maybe regulated those things to either happen more often or to happen quickly enough for a big company that you kind of have to get on the horn and then, and then of course you get it like into a whole pricing conversation, which is very, very different. so anyway, so anyway, so it's like the pricing is sort of a part of it. And then the flip side I guess is just people. Wanting to own their

In,

or not think about it. And, and, and there's more answers. Maybe I've talked for too long and you probably have an interesting question, so I'll, I'll stop you after that.

But I

I think

we've definitely also just in Canada, because for like three seconds, Amazon was like, I. Kind of like the big enemy here for a second. Um, they fired like a ton of people and it was like a really big thing. So a lot of people kind of turned on them. But it was interesting 'cause the

did

kind of flipped to like, well, like we did host all of our data on a weird American server.

Like everyone also just kind of had this moment of like, yeah, I get like, you know, like, and I'm not saying like people didn't realize that before, but I know I just talked to somebody who's like in the healthcare world. And he's like, yeah, of course our data's hosted in a data center in Quebec. Like, what are you talking about?

Like, of course our da we know where our data is. Those things kind of matter. And so the, the, the, the, the use cases where you were telling me about this in, uh, um, insurance company that is like very aware that you, you don't just go

Amazon,

you don't think about it. That's just like, you can't not

not

where your infra is or what

what.

Is.

Um, these thoughts are kind of like, it's a. Whatever. Yeah. Also like the, referring to that insurance company you're talking about, uh, that we talked about just before the call, um, so that someone I know where they use many clouds like, um. Google Cloud and Amazon and, uh, many providers like that. And they tried to switch to only Amazon to get a deal, basically because mm-hmm.

Amazon does that when you spend like, uh, 20 millions a year. Uh, and they did that, but then like it still cost a lot of money to, uh, to go on the cloud compared to if they moved back to OnPrem on-premise. Like they were like, uh. 10, 10 years ago, for example. Yeah. So like the, the big move to the cloud for big companies, uh, it was sold as only magic, but now mm-hmm.

The magic comes with, uh, an invoice. And a lot of companies are starting to think that yeah, maybe they could move some or all of their infrastructure back to their own data center. Um. Uh, so there's a need for everything, basically what I'm saying. Yeah. Not saying that, uh, we should like all be on premise and have our own servers, but I think there's a, for some, some organizations it could make sense to be on premise.

Some other organizations be on AWS, some other organizations be like on, uh, platforms, like on Heroku that does everything, or Versal that kind of hides, uh, all of that for you. Yeah. Yeah. Yeah. I, I think the mix and match is sort of like, it's, it's interesting. Yeah. Yeah. For sure. I think that's kinda the answer.

Maybe.

Andrew: Yeah, when I was looking at your guys' docs, it really reminded me of like early Zeit with their now project. Like kind of the same, similar like very simple JSON one command deploy everything. Uh, I love how it's like.

[00:42:59] Raspberry Pi Hosting

Andrew: As you said, you can have like whatever infrastructure you want, like even on your website, you're like, here's how to host it on a raspberry pie and the hundred steps that you would need to take to do that.

Greg and Antoine: Yeah, which I like. This is where, um, I don't know if it's the, the pilot or the reasonable or just me being like, really, really, um, some days having just like really wanting to, the freelancer brain want, wanting to do a new thing. I, I am like, I. I know we could make that Raspberry Pi experience so magical that it could be like an SD card that you get from us.

You plug it in and your thing is like fully disco ified and it's great. And Antoine, I, and I can even feel, he like wants to like, again intervene. 'cause he has to, he's done that many times. I'm not announcing anything that we're gonna do that and we're announcing it. No, but uh, imagine, um, the truth is yeah, the raspberry pies are. Um, are a lot of work and it's sort of like they, it's not that they're not our focus. We kind of like, we adore that it works and we have, like, I'm staring at two right now that are hosting a couple of things. Um, but it is like, yeah. There's also like the reality, like I've, I've had a couple of recurs. And I've already gone through like SD cards and at some point on one was like, we're gonna use an SSD here and like we are using an SSD uh, there. So there was a Raspberry Pi at Reers that's hosting like around 50 people's projects. It's something that we set up like way back and it's been super, super fun for people to like try it out and it's just like, you know, and it was like one eight gig, eight gig ram raspberry pie running 50 things.

So it's a lot of like dog footing and kind of having fun. It was great. But um. You know, it runs an SSD, which is like quite a challenge to put in. I once bought an SSD for new Rasberry Pie and it came with two cables labeled version one, version two. They looked identical and I was like, surely that's like a joke.

And I plugged the Ron cable and then the thing booted and then the errors were like, really? You know, those like deep, you didn't install the right, like, it was like really cool to see. 'cause that really brought me back to like, I had like really super scale. Like am I corrupting data? What am I doing? Then I had to replace the cable.

The other cable worked way better and I was like, yeah, this is what people do for a living. Like I'm not gonna like, you know, this, this is a little bit past the line of like, run your thing with us. I'm like, maybe not to that extent where, you know, obviously like if there's a storm, your thing goes down.

So Raspberry PIs, I'm again announcing it's like. It's fun. It's, it's almost like it's perfect for tinkering. And I think almost to answer why there's a hundred steps, this is really like a sour grapes kind of answer. But like, it's almost like you have to go through a hundred steps. Like you have to, it has to be worth it.

Like, you know, once you go, you really feel good about yourself. That's how it should feel like. 'cause at the end you should be like, yeah, I did a hell. A lot of things here. Uh, it's almost, we don't wanna make it too simple, but in reality we would wanna make it really simpler. Um. not right now. Uh, join the team.

Um, if you'll work on that. Um, um, um, at the same time, it's surprising the amount of, uh, services you can host on the Raspberry Party. Like it's runs and like. Yeah. Uh, like you said, Africa's like, there's like 50 projects and Yeah. I think backend projects like, yeah. Backend projects. And it's still using like, maybe like one of the gigs.

Yeah. So there could be So many more projects on a single small machine like that. Yeah. But still it's for like things like, I have my personal websites on my Raspberry Pi at home, and like it's connected to a backup, like battery, uh mm-hmm. Backup. And my modem as well is on a battery backup like that. So like if there's a power outage, it's still working.

Couple hours. Yeah. But still, like, you never know, like you're not home for a few weeks and then there's something going up and like, it's not like a production server you can run at home like that. It's more for fun and like for like projects that you would have like a. Paid the $300 a month. Yeah, yeah, yeah.

Perfect for me. Absolutely. Yeah. Yeah, yeah, yeah. Um, did that answer the question? I feel like we, again, went off on so many tangents.

Andrew: Uh, I, I,

I think we got to it

Greg and Antoine: Yeah. You sure? Yeah. Okay. I feel like, if, and I'm not complaining, I think if the an

yeah.

Yeah. I, it's like, yeah, we hear a word and we're like, oh yeah. That time that I, and then, sorry.

Yeah. This is like, uh.

We're just very excited to be here. So, uh, is this a good time for an ad read? Should we like do an ad read or do a thing? Do you

Andrew: Oh, we cut those in afterward.

Greg and Antoine: okay. After

Andrew: Yeah.

Yeah.

Greg and Antoine: did you see

the, did you see, you know, there's like a guy who, like, he, he like scams. Scammers.

Like he changes, he's like super popular, right?

Yeah. uh, he's like huge now. And he does these like fake coin base things where like the scammer things that are in the real Coinbase. And it's incredible. Now

he's like partnered with Coinbase or some other like crypto thing. at some point he got the scammer to read lines that were like, 'cause he was like pretending to be scams.

So he was like, oh, is scam, is Coinbase good? and then the scammer was like, oh, yeah, Coinbase is really reliable. And Coinbase was like a sponsor and he got the scammer to do the ad read. And I was

like that's. No one

will top that. So if you ever wanted us to somehow make an ad read without, I'm just,

I'm just, I'm just opening the door to like something that we could do

here, but we don't have to do that, but

I'm just really excited.

Justin: You already did a did a

Greg and Antoine: Okay. Okay.

Was it Okay. Was it Coinbase?

Justin: you did a, mux, you did a muck.com spiel earlier, and we had mucks as a sponsor in the past, so

that.

Greg and Antoine: Oh my God. Yeah. No. Oh my God. Yeah. No, this was, no, no. And

it's so like, yeah, it's from the heart. Mux is like, yeah. Seriously. I am not paid to say how much I love them. I really do. Okay. That's great.

Justin: so you're working on Disco Cloud. Uh, so it's a managed, uh, service, uh, for deploying disco apps. Um.

[00:48:09] Open Source and Business Models

Justin: Uh, and disco itself is an open source project. So there's this fundamental tension that you have when you're building an open source project and you're wanting people to use it and do it themselves, and then you're like, oh, or you can pay us to do it.

Uh, there's this like, usability question is like, if you make it incredibly, incredibly, like trivial for someone to do it themselves and it's significantly cheaper than what you would charge them to host it for them, uh, how do you balance that? Make it easy to gain adoption without like undercutting your sort of like value add on your hosted platform.

What does that look like for y'all?

Greg and Antoine: Yeah, it's a huge conversation. I think it touches on all of these like stories and topics of. You know, projects.

having one license and then realizing that it was too easy to use them. Um, I'm kind of thinking like, mad Box was an interesting example, right? Where like the library was like quite open source and it turned out that their business model, uh, I'm not saying it was just that, but a, a part of their business model was to host things on an S3 bucket for you. Uh, people figured that out pretty quickly. And so, Uh, if you hosted the, the kind of vector data, which itself was from open source open Streete map, um. Uh, you could not use the part that Mapbox was hoping to use to make money, uh, and

use their library, which is incredible. And the Mapbox didn't like that.

And then at some point they fully changed the license and and actually it's, you, you are not allowed to use the Mapbox library today, the JavaScript code, uh, if you're not hosting it without, if you're not giving them money. So it's like, no. What whatever version of open source is, that's what they're doing right now. and of course there was a big fork and it was map lever and everything. Mapbox are great, but you know, that was a big change. So I, I'm, I'm just kind of bringing it up 'cause it is things that we've like all kind of like bounce each other on. Also, it was like, oh, are, are we them? like what version of bad guys we wanna be or what version of like, guys we want to be in there.

You know, like we think about Mongo as, like, Mongo seems to have a business running the Mongo cloud. Um. I know of a gigantic customer. Am I allowed to say it's just, you know them like they're that big? A hundred percent. Everybody knows who they are and they are like a Mongo customer, which I'm like, how many numbers is that?

Like just probably nine numbers, like worth of dollars per month. Um, and it also makes sense. Yeah, you probably don't want to be like running like the core, core, core, core, core database or it's really nice that a whole other. Team of a thousand other developers are like really, really worried about your data.

Not like expiring and dying. Um, so that's interesting. But Mongo still, as far as I know, exists as an open source thing. Um, and it's interesting, like when you log into the Mongo, uh, uh, cloud, there's these moments where they're like, oh, like are using Mongo? And it's like, yeah, I might using Mongo, but are using Mongo plus Atlas Premium.

And you're like, no, but I'm, I'm not giving you money. It's like, ah, no problem. But you could be doing AI and we could be optimizing. You're like, oh, that, that's cool. That's, you found another cool way of making money. So where are we in there? Uh, I'm not trying to not answer by talking about everybody else here.

Uh, I promise. think that we, you know, now, like what we're trying to do is that we, we, care with, you know, our MIT license, we care about everything that we're working on, being open source and being available. We, we care that people use it and like, it, about the, not just dog fooding, but the, the thing that like, it, it is in the open like we are in this business because we are on, you know, on so many, um, you know, so many people before us worked and contributed and, and opened their code and it's, it's like. I, I think that makes a lot of sense. Um. We want that. What I think we're trying to go for is to make just it even easier for people who have more of, again, this sort of like what the commercial needs feeds feels like. It's sort of like, could we just trace the exact line where we're not inconveniencing free or open source users because we want them to, to give us money, but rather we're just exactly. We drew the line where people were about to give us money anyway, like a company. I think, you know, companies that like rely on their infra and don't necessarily have like their whole kind of, you know, like, uh, hosting staff kind of on board and DevOps. They might just be like, kind of like, they might not be interested in this thing being so open source that they don't even have a way of. You know, outside of some like sponsorship that's kind of pretty complicated to put in on, on place and you know, you have to be kind of sequel light to get a, to get a level where like, you know, like Apple's giving you like a million dollars 'cause it's on every funnel. So like, that's just like, we're not there obviously.

So I think right now we're kind of like drawing that line, but it's very. You, you know, I think we would just kind of be sad with ourselves. This is kind of like, you know, and I hope this, not that this haunts me, but like, please feel free to clip it and show it in my face in 50 years if I'm wearing like a weird hat and I'm like, ha, ha, ha. thought you could deploy? No. Your third deploy, like, you know, it's like just, you know, charges your credit card. It's like, I kind of feel like why would we make it more annoying for users to use? And, and, and clearly like why would we tie a, a business model to a thing that's all about like automation and making it simple. Um, so yeah, it was kind of, you know, yes. It's like, it's almost boring how we're like, went, we're like, I guess we'll just host disco and charge for that. Like, in a way it's kind of like, I. It's almost too obvious, but it does feel good in the sense that it doesn't feel like what we're doing kind of doesn't have anything to do with the software.

Right. It's like e even looking at like what happened with WordPress and WP Engine in a way, that's exactly what happened. That was almost funny for me to read. I'm like. WordPress was open source. And then, you know, and I'm not saying like the whole story, just like, you know, is like really just summarizes to this like, single thing.

Like of course this fight is like, again, like so many things happened, but it is be like, yeah, those other guys were just hosting your code, like hosting, like they offered hosting, your code was open. What are we talking about? And so in a way I'm like, you know, here's like a, you know, maybe I can promise or, or you know, challenge future Greg to like. I nev I would be happy if somebody offered a commercial disco cloud hosting and, and then you might be out. But what about Elasticsearch? They didn't like when Amazon did that to them? And, and then, you know, it's, so, it's like we're figuring it out. But, um, we went with that again because it's like, I don't think we wanted to just launch a thing that you could kind of like never see or tinker with.

I think we're still too. We just still like code too much to, you know, I, I would, I would, I think I would personally, even though Antoine wrote most of the code, I would still rather be remembered for my, you know, GitHub copilot generated, you know, code in the command line tool, which is where I usually write the code.

And maybe I, I don't know, I, I won't speak for Antoine, but I, I would rather me remember for that. And then for somebody to be like, yeah, like, it seems really good. Like, if I If I get run over by a bus, but somebody was like, disco, disco sure works well, things passed dead, Greg, that feels way better as a tribute than like, Greg sure has my credit card number.

Like, is that this, will anyone say that at my funeral with like, you know, with a little tear. Uh, but I can point to people who are hosting their things on disco and they, they seem to be happy and I, I, I. that's, that's kind of enough in a certain way, that, that feels enough of a whole. Is that good?

Yeah. Also would add maybe that, uh, for paying customers like we are, uh, there for support and also we are like, uh, hearing what their needs are. So, uh, there are a few things that, uh. Can't be done without an external service. For example, like if they want like a, team management or if they want like a enterprise or signon, things like that.

Yeah. Sam or whatever. Yeah. Yeah. So the things like that, uh, we could provide through, uh, paid, uh, subscription, things like that. But it would be harder to implement, like in a, in a demon for example, that would, like, if, if for example, you want to control many machines and like have teams that can access only a set of machines, things like that, then like it makes more sense to be hosted somewhere else.

And that's somewhere else I feel like is the, the platform that we, uh, we charge for. that's one thing. And yeah. And the F side, uh, of the open source, uh, thing is, yeah, like, Uh, Uh, Greg mentioned like the, of like, uh, big, big names, uh, starting to host for, uh, your code because it's open source.

Then you're kind of, uh, you know, like I, I feel like we, we talked about this a lot, uh, Greg and I, and, um, I. still feel like it, it's a, it's a great choice to have went with an MIT license for the, the, yeah, and we definitely want to make it, uh, as easy as possible to use. And for, for me it's definitely like the, uh, the way I see it, if, uh, we had like a million developer using the open source version for their, uh, AVI project. like it would just make sense for them

when they work in their day job to when they want to host a project somewhere, they would probably want to use disco

as well, because they're used to that tool. And then, uh, because then in that context, they would've the need for, uh, enterprise features like, uh, team management and things, things like that. Then it would make sense to pay in that context, but not for the average projects where they host it on

either

small servers at $5 a month or on the raspberry buy at home for free. Yeah. Or what is it?

Soc two, Soc

three, all the

socks. Like, you know, like that's harder to guarantee, but might care about that a lot.

And, and so that, that just again, feels like we're not screwing somebody who like just really wanted to run a restaurant rip kind of website.

Does that answer the question?

Andrew: I think so, uh, that that does

wrap it up for our questions and our time here. Uh,

Yeah.

[00:57:30] Conclusion and Future Vision

Andrew: for coming on the podcast. This was a really fun, uh, jaunt into all things hosting, and I hope we're all just hosting everything on raspberry pies in the future using disco. That that would be a great future to have. So thanks for coming on.

Greg and Antoine: Awesome. Hey, thank you so much. Really appreciate it. Bye.

Justin: Yeah. Thanks. We, we appreciate having y'all. Uh, sorry about my network connections. It's been really rough today,

Greg and Antoine: Oh, no.

Yeah, no, no worries. no worries.

Justin: Yeah, disco looks awesome. Excited about the self hosting future. Uh, really like your site design too. Uh, it's very like Cleany Classic. Love it. Uh, so folks are interested in checking out disco.cloud.

You can check it out for yourself.

Greg and Antoine: Thank you.

​