Epic Web Conf '24 Speakers

Glauber Costa

Founder/CEO of Turso

Watch Talk


  • Thursday, 1:50 PM1:55 PM
    Seriously SQLite


Glauber is an industry veteran with decades of experience in low-level systems on his back. His hair would be gray at this point if he had any left.

After an array of high impact contributions to key infrastructure open source projects like the Linux Kernel and ScyllaDB, he is now the founder and CEO of Turso


share on twitter


Hey Glauber, why don't you tell us about yourself? We'll get everybody to get to know you. Awesome. First, thank you so much for having me. I'm super excited for the conference. I just want to repeat what I said on Twitter. It's not going to be a good conference. This is lining up to be an epic conference, so I'm happy to be there.

I just got my tickets today, so I'm super excited. I'm Glauber, calling in today from London, Canada, not the one in the UK. I moved out of Toronto a couple of years ago to this small town called London,

two hours west of Toronto, and now I spend most of my time explaining to people that no, I did not move to the UK, even though my LinkedIn profile clearly says London, Canada. So here I am, excited to fly to Salt Lake City soon.

I am the founder and CEO of Turso, and I come from a systems programming background. My first dabbling into tech, let's call it that way, was through open source with the Linux kernel in the early 2000s, where I worked for around 10 years, most of the time at Red Hat,

some of the time in another smaller company, and some of the time purely in the beginning, purely as an open source contributor, which is how I started. I later joined a startup that pivoted into a product called Scilla,

which is a database company in the NoSQL space, and that's how I got involved with databases. And my co-founder and I, we've been huge fans of SQLite, as I know you are for a long time. We believe that it's a very underappreciated piece of technology,

and, in fact, the title of my talk is going to be SQLite is not a toy, as lots of people seem to believe, and I'm hoping to be able to convince a large part of your audience about that. And Turso is just like SQLite for production. That's how we talk about it.

That is awesome. Yeah, 100% agree, SQLite is not a toy. I'm using it in production and getting many queries on every request. I should probably record it for sure, but each request, when you go to kentcdots.com,

it hits the database with a bunch of different queries. Every single page is dynamic, all loaded from SQLite. I actually have two databases, one for a cache and one for the actual data. And, yeah, it's screaming. It's so fast.

And there are so many good things about zero latency, and we could go on and on. And you run it in production. By the way, when I claim we are SQLite for production, I don't mean that there is no way to run SQLite in production except for us. I mean that we want to make that journey easier, right?

That's what we're trying to do with Turso. But the talk is not going to be too much about our product. Anybody is welcome to go to turso.tech and take a look at what we offer, take a look at what we have there. But, again, mostly about SQLite itself as a technology, the cool things it can do. Some of the limitations, like everything has limitations.

Like I think sometimes people go a lot into this mindset of thinking, oh, what is the best technology? The best, the best, the best. But the reality is that, like, everything has pros and cons. So I want to talk a little bit more about that, about, like, how, again, everything has pros and cons.

But I still believe that for a large variety of use cases, not a niche amount, for a large variety of use cases, the pros are much, much higher than the cons. Fine technology, you know, maybe I'll get a quote from you, Kent. Yeah.

Well, yeah, for what it's worth, I migrated from Postgres to SQLite. I am a super fan. And, like, in recent years, a lot of attention has been coming over to SQLite. So, of course, Terso, what you're all doing, Fly has LightFS and LightStream,

and then Cloudflare has their D1 data, which is SQLite-based. Even, like, I think IndexedDB or some storage thing in the browser is also SQLite-based or inspired, at least.

So, anyway, SQLite is fantastic, phenomenal technology. I'm super excited that you're going to be speaking about that database at Epic Web.

Is there any other, like, peek that you can give us into what you're planning on for your talk at this point? Not that I prepared the talk already, but I think that likely it's going to be done. You know, I've been having a fairly busy schedule.

We're going to have a very busy month of March on our side here. So I've been focusing a lot on that. But, like, I do have a more or less sketch of, you know, something approximating a mental model of what I want to talk about.

And I think it will just be talking about the limits of SQLite, right? And in talking about the limits, you can show that, yes, it has limits. But, hey, look, I mean, it's not small, right?

So, like, in talking about the limits, you can talk about what the database is capable of doing, right? Yeah. That's actually – I love that approach because I think that a lot of the time when you're a super fan of a particular technology,

people are always asking you, well, what are you going to – like, what will you not use this for? When I was working at Remix, people would constantly ask me, like, okay, so, like, Remix sounds cool. What is Remix not good for? And people are always very interested in what those limits are.

And I think what you're describing is really going to surprise people. They're going to hear things that you say about SQLite and its limits, and they're going to say, wait a second. I'm not even close to that limit in my use case.

And so maybe I should give SQLite a try for all the benefits that it offers. Yeah. And maybe this is inspired by a book that I was reading the other day that talked about, like, when a –

there's a detective that is, like, collecting evidence for a case. And then he talks about, like, how can they determine, like, that the evidence was tampered with and modified. So you see, the claim is not that the evidence was not tampered with, because all evidence at some point, I mean,

something will happen, and especially eyewitness testimony, which was one particular example he was describing. Like, over time, as you transcribe this and et cetera, like, you can have variations. Like, every time you tell the story, you can have a variation.

But if you have a track record of what those variations are, like, you don't have to go and try to defend this wasn't altered. Just say, no, we know how this was altered. And because of that, I have actually even more confidence now on the original version, because I know exactly –

like, in a sense, you have more confidence on the original now that you know which alterations happened. Then, like, if you are – I was going to say, now, this was – of course, this wasn't altered. This is the original testimony. And, look, think about it. It's the same reason why we have version control.

And, like, you can see how your software was evolved over time, and you can learn something from that. And I think this is – you know, as I said, I haven't sat down and written the talk yet, subject to change. But I think that the mental model that I plan to approach is this.

I mean, in showing you the limits of this technology – right? So it's not a claim of, this is great, it has no limits, and such. But, like, in showing you the boundaries, you can understand by yourself that what lies within the boundaries is a lot, right? Yeah, yeah, I love that approach, Glauber.

I think that people are going to love that. And, yeah, I really appreciate you coming to speak at the conference. Is there anything that you would like attendees to come up and chat with you about, ask you about when they're at the conference with you?

Look, I'm not an application developer or front-end person, but always eager to learn. So I would love to get people coming and telling me, like, what they want from their databases and what their use cases are. Like, what are the problems that you have in your day-to-day?

You know, how other databases, and especially databases like ours that are like SaaS offerings, do not provide you with the best experience, and how can we do to help? Again, I come from this background of, like, systems-level databases. I don't have a lot of experience writing application software.

So the way for me to understand how to provide you all with the best services is by talking, hearing, listening. So this is super valuable to me. And people can also come and talk to me about whatever they think is valuable to them as well. Like, let's not be selfish. So overall, if you haven't noticed through this interview, I love talking.

Just come and talk to me. We're going to have a good conversation. Super. Hey, thanks so much, Glauber. Really appreciate your time, and we'll see you in April. We'll see you there, Ken.