These SQLite + Turso Use Cases Make DB Management TOO Easy with Glauber Costa, CEO of Turso

Summary

In this episode of Modern Web Podcast, Glauber Costa, CEO of Turso, discusses the innovations and advantages of using SQLite and the managed database service, Turso. He explores why SQLite, often perceived as lightweight, is rapidly gaining popularity in modern architectures, while emphasizing Turso's role in enhancing SQLite's capabilities for developers. Glauber also delves into the transformative potential of having local databases on individual devices, ensuring fast and reliable access to data with minimal network dependency.

Highlights

• SQLite is becoming more popular due to its simplicity, often underrated because of the 'Lite' in its name. 🌟
• Turso offers a unique managed database service leveraging a fork of SQLite. 🔄
• The edge computing concept is misunderstood but crucial; Turso provides databases starting at just $5. 🌍
• With Turso, developers can manage multiple databases efficiently, catering to innovation. 🤖
• The local database model on devices can enhance speed and reliability, optimizing user experience. 📲

Key Takeaways

• SQLite's growing popularity is reshaping database management with its simplicity and efficiency. 📈
• Turso enhances SQLite, offering a managed, cost-effective service ideal for developers. 💡
• The local-first approach allows databases to run on client devices, reducing network latency. 📱
• Cost-effectiveness: Turso provides an affordable plan starting at just $5 per month. 💰
• Distinct architectural choices make Turso and SQLite a competitive option against traditional databases. 🚀

Overview

Glauber Costa, the CEO of Turso, guides us through the intricacies of SQLite and how Turso is redefining database management. He highlights how SQLite's 'lightweight' reputation might be misleading, as it offers powerful tools for developers to manage databases efficiently.

In an engaging discussion, Costa explains the edge computing misunderstanding and introduces Turso's affordable, reliable database service. Turso, starting at just $5 per month, allows developers to create numerous databases without worrying about traditional database management complexities.

Finally, one of the standout features of Turso is its ability to sync local databases across devices. This innovation drastically reduces network latency, allowing applications to function seamlessly even in offline scenarios. Costa envisions a future where local databases on devices become the norm, enhancing overall application performance.

Chapters

• 00:00 - 01:00: Introduction and Guest Introductions This chapter begins with a discussion on the speed of SQLite and Turo, focusing on their efficiency due to local and mobile device operations that minimize network time. The conversation then transitions into the formal introduction of the podcast 'Modern Web,' hosted by Danny Thompson, Director of Technology at St Labs, along with a co-host.
• 01:00 - 03:00: Guest Background and Career Path The chapter features a conversation with Glabber Costa, the founder and CEO of Turo. Adam and the host express enthusiasm for having Glabber on the show. Although Glabber and Adam are meeting for the first time, the host mentions having engaged in several meaningful conversations with Glabber in the past. In this chapter, listeners can expect to learn about Glabber's background and career path as he introduces himself.
• 03:00 - 05:00: Experience at Red Hat and Linux Kernel Contributions The chapter discusses the speaker's early career and experiences in the tech industry. They started around the early 2000s, contributing to the Linux Kernel, which eventually led to a position at Red Hat. This entry-level experience in significant tech projects and open source communities highlights the importance of starting strong in one's career in tech. The conversation is framed around the backdrop of a tech conference, emphasizing the importance of networking and conferences in the tech community.
• 05:00 - 08:00: SQLite and Turso Overview The chapter provides a brief career overview of the speaker, who has worked at Red Hat, and engaged in other gigs, spending almost a decade in Kel. The speaker then spent eight years at a database startup called Sila, followed by a short one-year stint at Data Dog, and eventually found a position at Turso. Overall, the speaker emphasizes a steady career path without much job-hopping.
• 08:00 - 11:00: Why SQLite is Gaining Popularity The speaker discusses their past work related to virtualization in the Linux kernel, noting the impact of their contributions. They also humorously mention meeting their co-founder and their playful animosity.
• 11:00 - 13:30: Business Model and Pricing of Turso The chapter discusses the business model and pricing strategy of Turso. It starts with an anecdote about how one of the early customers of Red Hat, a bookstore then known as Amazon, developed the innovative idea of renting virtual machines. This involves delivering computing resources elastically, responding to demand surges. At the time, the speaker reflects on the nascent stage of cloud computing services and how it laid the groundwork for what Amazon Web Services (AWS) would evolve into, although the specific services like S3 might not have existed yet. The chapter likely uses this story to draw parallels or contrasts with how Turso approaches its own business model and pricing strategy.
• 13:30 - 17:00: Unique Use Cases and Benefits of SQLite The chapter discusses the speaker's experience working with virtualization and various aspects of the operating system. The speaker expresses an initial desire to work on kernel hacking but ends up exploring multiple areas such as the x86 boot sequence and virtualization, which touches different components like scheduling, memory management, and device handling. This exploration allowed the speaker to engage with multiple parts of the kernel.
• 17:00 - 21:00: Comparison with NoSQL and Future of SQL A discussion on the core elements of storage networking, focusing on memory and processor aspects, specifically x86 architecture. The speaker mentions interactions at Epic camp and their efforts to understand more about sqlite, traditionally viewed as a lightweight database.
• 21:00 - 27:00: Addressing Security Concerns with Local Databases The chapter discusses the growing popularity of local databases, particularly SQLite, in modern architectural setups. It touches on potential misconceptions about SQLite due to its name and explores reasons behind its recent growth and adoption.
• 27:00 - 34:30: Closing Remarks and Social Media Links The chapter provides closing remarks and social media links. The speaker reflects on a sentiment that leaves a lasting impression, suggesting it's hard to overlook. The speaker takes a moment to introduce themselves as the founder and CEO of Turo, explaining that Turo offers a managed database service stemming from a fork of Site. The chapter recaps the company's origin story, mentioning that the speaker and their co-founder were initially pursuing different ventures.

These SQLite + Turso Use Cases Make DB Management TOO Easy with Glauber Costa, CEO of Turso Transcription

• 00:00 - 00:30 so we've been talking about how SQL light is so fast and how turo is as well but like what makes it so fast I'm still trying to hit to the core of that look most of the time when people are talking fast and slow most of the time in inquiring databases is Network time seite can run locally right it can run on your mobile device and then the turo service keeps that in sync hello everyone and welcome to Modern web podcast I'm your host Danny Thompson I'm the director of Technology at this St labs and I am joined with my co-host
• 00:30 - 01:00 today Adam rakus Adam how you doing wonderful thank you awesome awesome awesome always happy to have you here and we are joined with a very special guest someone that I've met a while ago and we've been having some really great conversations and I wanted to bring him on he is the founder and CEO of turo and I am joined with glabber Costa glabber how you doing why don't you to introduce yourself for those that don't know who you are doing fantastic man thanks for having me thank you Adam for having me here uh Adams I think it's the first time we we're chatting so I'm looking forward for a
• 01:00 - 01:30 great conversation D we met at the Epic Web Camp uh what a great conference what a great conference that was right so I I uh I'll keep it simple I I'm a technologist for 20 something years I started my career uh around the 2001 bubble uh something like that and a couple of years later I I was contributing to the Linux Caro which is how I started I then joined uh Red Hat uh at the time uh and worked in most of
• 01:30 - 02:00 the time for red hat but like had another gigs as well in in still in the Kel for almost a decade uh I have then uh worked for another eight years at a database startup uh which is how I got acquainted to databases called Sila uh and after a very short unfortunately stint at data dog that only lasted one year I found a torso yeah so that's me in a nutshell I don't I don't job hop too much never did uh so my my career is pretty simple to explain
• 02:00 - 02:30 we got a lot to get into before we jump into the turo stuff can you just give us a little blurb on what you did in the Linux Colonel because that that's pretty cool oh man I've done all sorts of things that well the the worst of them was meeting my co-founder so like uh we we pretend that we hate each other but it's a it's not too hard to pretend the guy but but like aeka and I we both worked in the konel and I I worked I think the the things that I've done that were impactful the most were around virtualization
• 02:30 - 03:00 uh one of our customers one of our customers at Red Hat was this book uh store called Amazon at the time and they were planning to start renting some of their machines using virtualization for this crazy idea that they had that you could just come and elastically like get a box uh called Amazon uh I don't I don't even remember if S3 existed uh but like a I think they did like a the actual VMS first uh I didn't pay a lot of attention to to any of that at
• 03:00 - 03:30 the time for me it was like I just want to you know be here hacking in the kernel but I've done I've done a lot of like lowlevel stuff in the x86 boot sequence of true virtual through the thing about virtualization is that it can touch multiple areas uh of of the operating system so in the time that I was working uh with virtualization I ended up doing a little bit of scheduling a little bit of memory management and you know a little bit of uh devices so I've touched multiple multiple parts of the kernel uh a little
• 03:30 - 04:00 bit of storage networking as well but but the core I think the core was uh you know memory and and pro processor specific called like x86 stuff thank you very cool well I'll ask you this you know we talked quite a bit at Epic camp and I have been diving through the docks quite a bit and so I was like trying to understand more and one thing that at least for me sqlite has traditionally been seen like as a lightweight database it's you know and
• 04:00 - 04:30 now especially like I find it it's been around for a while but it's getting a lot of traction more lately than ever and so I know with you that's definitely an area where you have some invested interest in it so I'm like what do you think is driving its recent growth and adoption and like why are people utilizing in their modern architecture setups like why why is it there well I saw somebody on X the other day saying and and I agree with the sqlite would have a lot more users if it didn't have light in the name uh I definitely agree
• 04:30 - 05:00 with that sentiment right just because it gives you this impression and and the impression is just hard to to to go around uh first uh just just for context to if I if I may take a quick step back just to say like I'm the founder CE of turo and what is torso torso is a managed databased offering that is based on a fork of site that we have right so that's a little bit of a mouthful uh but it all started uh we founded this company my co-founder and I we were doing something else entirely we were
• 05:00 - 05:30 using cite we were users of cite in our product uh we found a couple of limitations with cite uh again we are long-term veterans in open source and and and low-level systems programming uh we knew we could fix those issues some of them actually very easily uh but the thing one one thing about the seoli project is that they very famously and publicly do not accept contributions they have uh they have this on their website isay like the cite is an open
• 05:30 - 06:00 source but not open contribution project uh and they actually say feel free to send us your ideas you might even send us code but don't feel offended if we accept your idea but decide to completely rewrite your code uh we just not going to accept your code and so so it's a it's it's a project that doesn't have a community of developer which is by by the way just because it's so easy to misunderstand things on X uh and in podcast we always have the chance to to to go a little bit deeper I just want to
• 06:00 - 06:30 make it 100% clear that I don't think this is fundamentally wrong right when we talk about it some people say oh they're criticizing cite they're they're dunking on cite or no uh you know they I think they they have all the right in the world to do it that way they say themselves if you don't like it Fork it then that's what we did like we figure like we want a version of this that has a community and we would like to fix a couple of those issues that we we we saw when when I say fix you may think it's a it's not a bug architectural things like
• 06:30 - 07:00 that we would like to do differently uh using the cite model so we forked cite we saw like this project getting like a 1500 GitHub stars in in in a week and a half uh something like this so it was a clear for us very clearly there was something there uh and then we decided to completely pivot our company in that direction and build a manage service around it right because it's one of the ways uh in in which in today's day and age you can make money like very hard to make money if if you're just like a fork off a file base database uh so turo
• 07:00 - 07:30 again is a managed service based on that and we offer a lot of interesting uh Services cloud-based services on top of that uh so this is where the our interest on cite comes from because you know ultimately uh the more cite gets adoption uh our Fork is fully compatible with cite of course in fact we have we make very few changes to that uh but the more adoption cite gets the more developers uh a few comfortable using the better it is for us there is there in lies the the business interest interest uh now with with that
• 07:30 - 08:00 introduction uh that was just contextual for uh what you know sqlite going into uh SQL why why is it seeing a lot of adoption now I think I think it's true fold I think it's true fold and uh the first the first one and for context as I said I spent eight years working at a nosql database company it's funny because I was in the Dax podcast uh more than a year ago and he he said I had a very he told me that so this is Dak
• 08:00 - 08:30 saying I had I had an unusual career path I learned no SQL before I learned SQL and I said dude that's nothing compared to me I wrote a noo database before I learned any of them because I mean I came from the Linux Kel and then I joined this company that started writing a noo database so that's where I learned my world was all this like I getting syllabus of database that could do millions of Rights per second uh would survive any kind of things like notes could fail at any time it would rebalance the cluster and stuff like that so this very like
• 08:30 - 09:00 highlevel uh uh operations to sqlite which is this tiny thing and the reason the reason I mentioned this is because when I go back to whatever 200 I'm getting old 2012 or or what not 10 almost everybody needed an nol database in in the sense that if you wanted to do anything at at scale you needed a sqlite database and scale at the time just meant that it's a website that I need to keep running right uh so
• 09:00 - 09:30 so if you look at [ __ ] for example [ __ ] grew a lot on that in the sense and why was that because like you barely had multi-core machines like machines were not it didn't have multiple cores as we have today uh the discs was not very big so if you had something like man I got like a 100 gigabytes of data you needed no SQL kind of scale like you you couldn't fit this on a box now what happen is that the the workloads didn't grow as fast like if if you needed 100
• 09:30 - 10:00 gigabytes for the workload or 10 gigabytes or whatnot in 2010 you'd probably need the same today or your shopping cart did you know your product catalog did not grow as much uh but the machines grew tremendously so that need to have like clusters of four or five machines to do the simple things does not exist anymore so so with that I think we saw a Resurgence in SQL forget SQL life for a moment it's SQL that is why for example a lot L of people today talk about post right because you can if
• 10:00 - 10:30 if you go back a couple of years in the past uh you needed post just wasn't enough for a lot of serious use case now it is it's postly is myo had a Resurgence so that that's number one uh and and and cite in particular I mean it's it's an incredible simp incredibly simple technology I think developers like Simplicity uh there is this idea if like if I can solve my problem with a very simple tool why would I not use a very simp simple tool uh it allows a very
• 10:30 - 11:00 local uh it allows a very you know Pleasant local experience because you can develop with with just a file and there are a couple of use cases that are very unique to cite that that are possible uh they are very hard to do with other databases some of those are the ones that that we specialize at torso so there are true use torso is is again a database that that serves many purposes uh one of the things that we do very well before I go into the use cases where we I think we absolutely kill it uh one of the things that we do very well is that uh if you go to our pricing
• 11:00 - 11:30 page you will see our hobby plan uh our hobby plan for hobbies developers costs $5 a month uh I don't think there is any other database in the market that can meet that price points I mean site being very simple translates to you know this is not very complicated to run I can I can offer you thisit at a very low cost uh but the use cases that that we do very well like first of all creating a database file it creating a database on on torso is it means like creating a site file that that cost nothing in in
• 11:30 - 12:00 in time and money uh it only costs the storage they use so it's very easy to make a serverless database out of this you see all of those companies trying to do like serverless post is very complicated sqlite is very simple because the file is on disk so you already have segregation of computer and storage naturally the files on disk if you query that file you pay for that request if you don't carry that file that file is just sitting on disk somewhere right and and then you can create as many databases as you want I provision infrastructure for you and and
• 12:00 - 12:30 you can create unlimited databases on tour so there are a lot of use cases that benefit from that some of them a lot of the AI use cases benefit from that because you can have like AI models operating straight on your database and then you just create databases on the fly so you can you can get your your your your you can get Claude for example to write live queries against site files because creating databases is so easy and cheap and then you create them you do your queries you destroy them so we see a lot of use cases like that uh and we and we have the more interesting use case as well with sync uh because site
• 12:30 - 13:00 is a database that is designed to run on devices on on fridges on toasters on on whatever you want uh so you can run those databases on mobile devices you can run copies of those databases on your server and then our our service at turo keeps them in sync uh but those will be like offline right SL local first use cases whether or not naming in computers in our industry is always a problem because there's always something hidden in the name uh but however you want understand those use cases where
• 13:00 - 13:30 you want the data to be local to your device those are also use cases where we do very well that was really interesting uh well all of it was interesting but one thing really caught my ear um where you talk about how no SQL kind of blew up because of the built-in horizontal scaleability which was needed because boxes were just a lot less powerful 12 years ago um and now we've seen SQL sort of come back into a new sort of Renaissance long term
• 13:30 - 14:00 as that Trend continues do you see any place for no SQL in the world it'll still be around forever absolutely uh and it's just and it's just about like uh what I think what I think happened and Sil is still around and kicking and actually doing very well by the way uh what and one example I think that is going to be more familiar to the web developers in general which I presume is a lot of your audience is planet scale right uh so Planet scale is effectively a no database uh even even though it is SQL
• 14:00 - 14:30 right uh it doesn't support foreign keys and and things like that because it's designed for scale right and and what happens like you all saw last year like one of the things that planet SC did is that they got rid of their free tier so essentially they kicked the developers out uh and and say now now you don't have a free tier anymore which a lot of the companies like including our company are always fighting to get those free tier developers because some of those a lot of the a lot of those developers will never pay as a dime they're just cost and and we're fine with that like
• 14:30 - 15:00 and and but this is you know top of the fun of developers come developers like it they enjoy it they keep building on it uh Planet scale by doing this essentially they said we're not interested in those developers anymore uh which by the way it's they were criticized at the time and and and and I think there were Fair criticisms about the tone and and all of that but it's a move that I supported uh in the sense that if I were their Co I would have done the same move because Planet scale is wildly successful wildly successful with companies paying them seven figures and and and and above right so if if you
• 15:00 - 15:30 have this kind of customers and it's very hard to build a company for two customer personas at the same time if you have this kind of customers why do you care about the folks that going to pay you like 50 bucks a month right you don't care you just don't care and you shouldn't care and then if they top of the fun of top of the funel for a company spaying a million dollars a month do not come from from from like weend developers doing random stuff I I don't know if I can curse on the podcast or not but like a um and and this is the story of no secret because what happens is that there before like everybody
• 15:30 - 16:00 needed no SQL to do any kind of useful thing now only The Fringe of the use the use cases need no SEO but what happened exactly because those cases are Fringe they're very very very valuable uh so you have a lot fewer people doing it but they're paying a lot more money because they the use cases where you really need this are incredibly valuable uh and there're not used cases where you can screw up so it's used cases where you need you really need to have technology that is incredibly rock solid uh not the
• 16:00 - 16:30 other Technologies are not but like the Investments you make in in reli everything is is Downstream from the Investments that you make as as a company right so like you either you either prioritize like in the case of Sila you could do whatever whatever you want and I think planet scill is in the same way uh you can do whatever you want to that database that database is never going down because it's replicated five times across the globe and and Etc right it's just not the same kind of infrastructure we have a t so because if
• 16:30 - 17:00 I have that I can't charge you five bucks a month right so so it's a completely different part of the market and and I think what happens with noo is this noo goes to like fewer use cases but the use cases where noo is needed are so incredibly expensive that it's not really going anywhere this episode is brought to you by this. laabs the experts in modern web development and Consulting if you're looking to scale up your developer teams efficiently or need assistance in
• 17:00 - 17:30 turning your innovative ideas into reality with robust proofs of Concepts this. laabs has you covered this Labs doesn't just build software they build Partnerships they collaborate closely with your teams to understand your unique challenges and craft tailored solutions that drive Real Results join the many companies that have accelerated their growth and Innovation with the help of this. labs visit this. to discover how they can help you take your projects to the next level they've worked with customers such as Google Microsoft stripe Wikipedia and we can
• 17:30 - 18:00 work with you as well thank you to this familiar with SQL light what is what's the point of using SQL light over something like postgress or MySQL which are both completely open source completely free why do you host that on your shared servers that developers can sign up for over the others well you there needs to be a distinction there I think because like uh about when you say things like that it is free right it's note free it it's it's software that you
• 18:00 - 18:30 can use for free which is a completely different thing uh by the way I wrote an article and we can go into that later called databases will be free uh about like how we driving the cost of a database to zero it's all part of this discussion but all of this to say that the word free is a little bit of a dangerous word right because uh the softare being free matters very little because yes posters is free but very few people are using posters uh most people will go use a postr provider right
• 18:30 - 19:00 because databases database and the provider is not free because you're paying for the provider to run uh to do a lot of things on your behalf and it's the same with us uh right so the comparison cannot start there because cite is also free as much as postris in fact is even the license is more permissive because cite doesn't have a license cite is public domain software so and which is part of the reason why they they don't take contributors because they they as far as I understand understand and you know maybe I'm I misunderstood this but
• 19:00 - 19:30 as far as I understand it makes it easier for them to control like keeping things on the public domain uh so so yes it's free but like when you're going to run any database service you have this option you can run posters yourself you can run sqlite yourself you can run my SQL yourself but very few people do this and and the reason very few people do this is that like like just your database is a key part uh of your of of your stack and it's a part of stack that you usually as a web developer usually
• 19:30 - 20:00 as an application developer you are not an expert on uh and being an expert on a database does not mean understanding the query language I actually met a lot of application developers that are much much better than me uh which is not that hard uh in the query language because again I don't come from this background uh but managing a database means having backups making sure that the backups are safe being able to restore those backups and making sure that that the database can withstand the load and and and all things like that right so so this is just to draw the distinction between
• 20:00 - 20:30 free and not free like you can again you can use postr for free that's fine but that's not the real competition here because in practice you will not in practice you're going to be paying super base or or neon or n or whatever other planet scale or in the case of myo Planet scale or whatnot right so why would why would a developer use uh site uh versus postgis and why a developer would use torso versus one of those other services is they're two different stories so so umy so what I was trying to drive for is why
• 20:30 - 21:00 did you as the managed database provider why did you opt to use SQL light as an offering rather than postgress like super base does or MySQL like other people do yeah uh so so that that's a good clarification and we as I said I mean just there there are two kinds of answers and and one of my problems I keep hearing this is I'm too honest right uh I I can give you the beautiful answer the answer that's going to be quoted on Forbes uh on stuff like that although I'm too I'm too uh I'm too uh
• 21:00 - 21:30 late for Forbes 30 under 30 because I'm already over 30 and I also don't plan to go to prison uh so there's that but uh the the the answer that's going to go in my biography it's something like we identified the use cases and and we had this Vision right but the real answer as I said is that we kind of stumble upon it by accident because like we saw we saw this rise of cite we saw a lot of people paying attention to cite and we were we were using cite in our product
• 21:30 - 22:00 then we saw our Fork of cite essentially getting like tremendous growth almost overnight uh and as a company as a company founder you know the situation is a it's the game theory is more complicated there because yeah postris maybe postris is more popular but at the same time because it's more popular it has like five six different competitors uh and you know site is growing like crazy and and there are no competitors so like we have a chance to to do something very unique and very special here which is which is also very
• 22:00 - 22:30 enticing as an engineer because I am an engineer by by background right so we we can do something here very unique and very enticing uh so for us was this like and now once you have that okay now before we even start doing it what are the kind of things that we can do that are very good that this can be very good for uh but I think I think it has and those things exist and and they were not very they were not very hard to identify we started doing cite we started doing cite for the edge in the beginning our our landing page was just this squel on
• 22:30 - 23:00 the edge uh and then what we found out very quickly is that first of all uh there are maybe 10 JavaScript developers in the world that understand what Edge is uh versel made a tremendous mess in my opinion with my personal opinion on the concept of the edge we started hear hearing stuff like I don't like the edge I prefer no JS and and you know keeping my background in mind I was like what the hell are you talking about like just geographical concept and then you're
• 23:00 - 23:30 talking about no. JS like but we understood what happened over time and people in our Discord Community kept telling us I mean that's the beautiful that's the beautiful thing about building stuff you build whatever [ __ ] you build people come and tell you what you should be building instead if you're building the wrong things just go build that's my advice for Builders right so people come to our community and started saying this is great this is why we like torso right and part of the reason is this because I can offer you a database for five bucks right the other providers on post cannot uh but they kept telling
• 23:30 - 24:00 us if this is site why can you why do you only offer me three databases because that that was our free tier we offer three databases on the free tier it's just a site file give me a thousand right give me a thousand uh and then we just by talking to those people we understood that the thing that is really powerful about site is this you spin up databases at well and then there are many creative use cases that you can do with that so we've been talking about how sqlite is so fast and how turo is as well but like what makes makes it so fast I I'm still trying to hit to the
• 24:00 - 24:30 core of that look most of the time when people are talking fast and slow most of the time in inquiring databases is Network time cite can run locally right it can run on your mobile device and then the tal service keeps that In Sync O so you're doing like you're putting read replicas right on the developers machines we do yeah and and we started uh we we started with just doing this like it's on servers right if you have let's say you have an API
• 24:30 - 25:00 server and then we put a copy of the database into that server you don't need a cache anymore your architecture is simpler and microsc level query you can't get faster than that over time again what we heard from a lot of the people in our community is that's great but can I get this on my mobile device right because the user like I I have an application and then that data base goes straight into the mobile device over time people kept asking that's great you have a read replica on on my mobile device but now I want to write to this
• 25:00 - 25:30 database on my mobile device as well so this can keep working offline so we kept building the service based on those very unique features of site that were already there is it's just like people telling us where they want the service to go uh but what you have with tho today is essentially this you can have a local database in your mobile device and by the way uh one thing that we keep hearing and we'll do it likely this year uh we haven't done that yet just out of bandwidth but like a browser can we have this now running on the browser uh
• 25:30 - 26:00 hopefully now in 2025 you will be able to it can run on your mobile device it can run on your API service it works offline and all of that so fast at the end of the day it's because of that like it's not a benchmark about like how the database executes on from your API point of view like the network time dominates the query time so all you have to do to make something fast is get the database in whatever you are well hold on so you're saying if if
• 26:00 - 26:30 I hold on so you're saying if I have users using my application I can create a local version of the database and put it on their device but like yes why would I put the entire DB on there like how do I know what they're going to end up utilizing because obviously that's a lot of Records there's a lot of and then on top of that you're saying that you're syncing it but if you're syncing it in that regard like how do you I guess how do you keep the quality of the data accurate at that point like what what's
• 26:30 - 27:00 happening there I don't know if you're editor have you seen that Meme that the guy does it like that and says magic can do that just know just joking but like it's not it's not it's not that it's not magic anything like that it's just like look uh I think there's a lot of solutions today in the market that try to go after this problem how to have local data local First Data and Etc and we have a fundamental belief we have a fundamental belief that we acted on that's the belief that we had when you know the
• 27:00 - 27:30 thesis so to speak when building this solution which is a lot of those problems uh you can have Technical Solutions for those problems syncing engines what what not but that the real solution is architectural this is how we view this so we always sync the entire database at the database layer so there is a boundary there which is called the cly database that's a cite file you're going to have a copy of this database whatever you want that copy to be uh you're going to sync that copy back to the cloud and then the cloud will our
• 27:30 - 28:00 cloud service will distribute that to a thousand other clients so you do whatever you want this is all kept in sync but the fundamental unit is a database and then what we do is that if you remember that one of the strong points of cite which is the thing that people kept asking us when we created the service is how you can have a thousand a million files uh sitting there and and they don't use up resources because they're just files so what we get you to do is that you split the data into multiple databases in a way that makes sense and I'll give you
• 28:00 - 28:30 an example very shortly and then we sync that entire database to the user's device one example uh one example is exactly this if you have a if you have an application that is serving users the data about that user is the data about the other users is just not needed when that user is logged in right so like when that user is logged in all that user cares about is his or her data so what you do is that you put data per user in different cite databases and then and then you sync only that
• 28:30 - 29:00 database to the user and you can have databases with shared State too so you can have a few databases that have shared State a couple of megabytes and then you sync that database to the user device and and and then you encrypt that one separately because it's shared data or maybe it's public you know depends on on on what you're doing and and then you syn on top of that the user data another example if you're let's say you have a book service and those are real life examples that we see some of our customers doing like you can put uh one of the things that we added to cite for context is Vector search so turo our
• 29:00 - 29:30 Fork of cite can natively do Vector search so let's say you're writing a service that can query that can talk to books uh using using uh whatever local model you have you don't you have you don't have to sync now your database with all the books because each book is a different site file so you sync that book with all the embeddings with all the AI models uh and now the user can ask questions to about that book locally uh and then you can if you want to
• 29:30 - 30:00 record the interactions that this user had this you record this interactions on a different database and then you sync that database locally you sync that database individually so we do two things uh we we we allow you to sync entire databases but then we give you a million databases uh to to sync from uh so it becomes very effective and very economical uh to have uh those use cases where you just have a lot of databases that's really cool and maybe I'm
• 30:00 - 30:30 overthinking this right and if I am correct me and we'll edit this out so I look like a g on the podcast but my question is this my immediate thought is by having like a local version of the database on their device if the phone is compromised in one way shape or form how do we know we're not compromising data of other information that could be I guess a vulnerability for us by having it like how are we ensuring that it's safe at that point well you don't you
• 30:30 - 31:00 don't you don't sync data right that so that's why the database boundary is important if you have if you have a sync engine that is syncing data that that is unrelated uh and most sync engines essentially work as a cache because you know you have to get the next 100 gigabytes the next 10 megab you query a row and then it tries to make sure that the the materialization of the query is there or the next 10 gigabytes are there or things like that uh then you start
• 31:00 - 31:30 having those problems but for us is exactly this like it is upon you to partition your data into multiple databases and then the data that you sync is data that you consider safe to sync now you can encrypt those databases as well which is another thing that you do so you you can you can encrypt that data you can sync that database and encrypt the local file that gives you an extra level of security but you're not going to have unrelated data because usually that's the problem right security is is a big word it's it's not
• 31:30 - 32:00 a Boolean thing secure versus insecure it's all about the Trap model right one thing is that user I am a user I lost my phone somebody got access to my data because of that that's bad and and you can protect against it one way but that's a completely unrelated and and problem to uh the user lost their phone uh now those other five users got their data compromised they have nothing to do with that right so so the the latter problem you solve by essentially by the database boundary uh the former problem
• 32:00 - 32:30 you solve for example with with encryption uh but those that problem is a lot easier to solve already because databases are no database like you you have to have those measures on uh uh but and and once more just just to make sure that that I paint the whole picture to the audience like the the mobile thing is one of the things that we can do uh but we can also be sying that you want API server which is a lot safer because you you're still in control of the whole infrastructure right the the local read
• 32:30 - 33:00 it's mobile only and you're working on a wasm solution for in browser for web apps is that right yeah well yeah well we're not actively working I wish we were I mean we we plan to this is all about time and and priorities and Etc sure it's it's on the road map basically it is on the road map yeah very feature that a lot of people ask can we understand man the browser I actually I actually in December close to Christmas uh because I had a little bit of of time in the beginning of December uh I took a look at uh my myself and like okay how do we get this stuff to run on the
• 33:00 - 33:30 browser the browser is a weird land man uh you know I did not know this I I kind of knew on on a passing way because people mention and Etc it it is it is it is weird in in a lot of ways so I mean some I think some that's what we run for example in in in web containers uh if you if you're familiar with stack Blitz for example we have a version that just runs on stack Blitz and this is relatively easy for for what we do uh you know the browser has a little bit of a a couple of extra extra complications sure globber thank you so
• 33:30 - 34:00 much for being on this podcast with us you've given me so much i' I've been taking notes to research and and kind of dive in a little bit deeper I love this aspect of having like the local DB on the device and I really want to find out more about that and kind of discover what that incription even looks like I think that would be really interesting um thank you so much for hanging out with us I am Dan Thompson I'm D Thompson Dev everywhere on social media you can find me there Adam rakus is at Adam Rus everywhere as well Adam thank you so much for hanging out with me today man yeah globber Costa you can find him at
• 34:00 - 34:30 gcst on Twitter or you can find him at glomer on LinkedIn um and you can find him everywhere you also find links in the show notes as well and you can find links to turo to check that out as well and before we wrap up this episode I would like to give one extra special thank you to this. laabs your partner in web development and Consulting at this. laabs we specialize in empowering businesses to reach their full potential through technology we understand that in today's fast-paced Tech environment adaptability and efficiency are Cru crucial that's why we offer tailored Solutions designed to meet your unique
• 34:30 - 35:00 challenges and objectives and our commitment to Excellence and customer success sets us apart as a trusted adviser in the industry discover how this. laabs can transform your business visit this. to learn more about our services and how we can help you stay ahead in the competitive landscape thank you to this.l are sponsoring this episode and for the dedication to advancing the developer community and that is all from us this time y'all and we'll see you on the next one goodbye everybody