Zero Gas, Infinite Scale by Luis Schliesske

Transcript

it's beautiful okay so I'm here today to talk about zero gas infinite scale the bull case for app chains and how to build them um so yeah first show of hands who has heard of gelato Network before here nice quite a lot cool so actually uh I was here at East Dubai two years ago a year ago and two years ago I talked about our Automation Services we gelato built the first decentralized smart contract automation protocol that's like four years ago and uh back then I would talk to you about how pancake swap uh mea Dow and so on have used our Automation Services to uh automate the protocol upkeep uh I think last year talked about our servess

functions uh gelato web functions uh how B Finance urine and so on many other projects are using serverless environments to to get uh I know offchain payloads from one inch and others and automate their protocols based on this and and this year I'm talking to you about something quite different but it all sort of fits together quite beautifully which is the world of rollups rollup as a service and yeah I think it's quite fitting because you probably noticed that everybody and their grandmother is launching an L2 these days uh and I will talk a bit about why that is the case so I'm Louis uh I'm the founder of gelato Network and yeah let's just jump straight into it so first I'm going to take you down a small

trip down memory Lane and I'm going to talk about how web 2 scaled so at the start of web 2 basically you had local hosting this meant that you put your HTTP server onto an actual physical machine in your office at your home uh you would have to install software on that machine uh you know and that was your web server and obviously there were many issues with that that didn't scale because I mean first of all you had to buy that machine you have to set it up you had to maintain it and especially if you were like we're building a Global website um that machine was in your basement somewhere and someone on the other side of the the continent or Globe even would have very slow loading times trying to reach that machine right so very

unscalable from there we went to Shared hosting um I don't really remember this I think I'm a bit too young for that but some people might remember GE cities and dream host and this really was a massive step up from from local hosting so shared hosting providers would allow you to deploy your web app or website in their shared environment online so they were managing all of the hosting for you the servers for you and so on and yeah sounds pretty cool and uh the problem was still that um it was very constrained in scalability and very inflexible in usability because the shared hosting providers would usually have a have a framework um that you were constrained by so you would have to use their programming language it would be quite limiting sometimes in what you

could do and worst even worse they like every was everyone was deploying side by side to each other so so if another app had a huge unexpected Surge and traffic you would maybe suffer from that other application that has nothing to do with you yeah maybe you see where I'm going with this so what was the endg game for web 2 scaling Cloud VM is basically AWS Google Cloud you name it right so the end game was to have the hardware layer completely abstracted and virtualized in the data center and to allow developers to use VMS like Linux and so on uh virtual VMS in the cloud and that was like that's where we at today that's I mean infinitely scalable you know um you can bring any program language uh it's completely you know up to you how you

want to use this and uh you can you have autoscale you don't share the infrastructure with everyone else anymore per se so you can sort of just pay the cloud provider for how much compute you need it's even allows you to do that automatically with autoscale an elastic compute and so on so that's that's really great you know you can literally deploy um your web app now and uh if you design it correctly you can basically scale to Infinity you know and that's been proven by big tech companies like Facebook and so on before right so the the key takeaway here is that the internet is not one huge World computer actually it's a web of interet interconnected computers and web 2 scaled via

asynchronous services so most of these Services even your your own application the instances that you run they talk to each other usually we are synchronous asynchronous interfaces HTTP uh you know all of these protocols um yeah so that's basically asynchronous Services were very key to the scaling benefits here so let me try and tie that back to web 3 and and the claim here is that web 3 will and is basically scaling via asynchronous cloud services too so I just took you the trip down memory lane for web Web Two scaling and now I'm going to do the same for web 3 so in web two scaling we had physical servers in

the first stage right at your home this was basically Bitcoin back in the day in 2013 I guess and so on when when people tried to replicate Bitcoin there were application chain application app chain developers out there like mastercoin and many others who basically back then had to replicate what Bitcoin did which meant they sort of forked it like do chain and so on right uh and the this was very unscalable because you basically had to replicate the success of Bitcoin which meant you had to recruit a massive valid Data Network of Miners and so on you had to inflate your own coin to pay them and so on and make that all work and that basically wasn't scalable at all and that's why ethereum was invented right um we went from launching your application as a chain as

a proof of work chain to launching your application as a smart contract on top of ethereum and that was like insane insane Innovation and it was really great and it really helped this space uh get to where it is today and those many application developers to exist right but there are problems with etherum right we know that so again similar to dros and Geo cities application developers have to share this space this compute right and that means if some crazy nft happens or some Ico you your users might suddenly suffer huge degradation because gas is going through the roof the network is congested and so on and you know now your application suffers from something that you have nothing to do with and also similar to Shared hosting providers web ethereum is

very constraining right like evm solidity and so on that's very limiting you can't build a much you can't build very powerful applications with that um you know maybe a few could good examples would be Unis Swap and so on but outside of that it's very hard to build powerful applications um on top of this evm platform we know that evm itself has many shortcomings it's not paralyzable and so on um so so yeah that's basically not very scalable and that's why we had to invent rollups and rollups you could really liken them to to VMS on a AWS or something you can think of as rollup as a virtual blockchain it's a blockchain inside a blockchain sort of um and yeah

rollup as a service providers like gato basically are like AWS where you can spin up many instances of of a virtual ethereum op arbitrum whatever you want right your instances just like you would spin up instances of Linux and so on on AWS and and you have you can easily configure the VM as well like that's not being done much today but it will I think very soon happen that you won't just have vanilla op anymore like vanilla evm op or arbitrum actually arbitrum for example already has stylus and so on so so yeah you will see parallelized EVMS like Monet and so on technology being used you already today have rollups like Eclipse that use Solana VM but they settle on ethereum and and have da Celestia for da so really the developer can now come again

they can bring a powerful VM like I I mean cartesia for example they're working on on fraud proofs for Linux VM so maybe for the first time we we we can write blockchain applications in in a Linux environment which is like the Holy Grail uh and and yeah and and even more important uh you have a um sort of virtualized service like gelato rollup as a service where you can spin up as many as as these as you want and you can you know scale them to your needs um yeah so that's basically the end game for web 3 as well and we're not quite there yet but you know at gelato we've seen like I don't know 10 20 chains launch on our platform only this quarter and it's really happening like many many applications are moving away from Smart contracts on on a layer one or Layer Two

and they're going onto their own block space to to be able to have dedicated block space to not share it with the other applications anymore and to only have uh scaling needs when they actually have their own traffic they need to scale and not some random other applications traffic so um you can think of gelato as the web pre- Cloud we have built middleware Services automation account OB section relaying all the stuff Ser list in the past and now we're bundling all of these up together with your own dedicated VM and block space and I'm going to now go to some technical architecture it's very high level so it's not very technical to be honest but but yeah you can roughly think about it this way when you want to basically

create your own roll app you have the base layer that you can pick so Bas layer would normally be um you know uh the execution engine like and obviously ethereum itself but the execution engine like op G or arbitrum or polygon cdk uh the data availability layer the data publishing layer like igen da Celestia Avil just launched right we did an event with them here in Dubai uh and on top of that so that's like the block space and the the sequencing the compute on top of that you need a bunch of servers for your block space to to be useful right like so you need gelat relay serverless maybe if you want to talk to oracles like Redstone or pth they all use usually web three functions to get the data from offchain to onchain you need block explorers like ether scan or block

Scout for sure right bridging so you really have this sort of basic rollup St stack that we provide you with uh but it doesn't end there you really when you when you launch your application as a chain on a smart contract you need all these other services as well that before you took for granted right like when you launch a smart contract on polygon or on ethereum you you get the graph there it's already deployed you get ether scan there it's already deployed and so on right here when you when you enter your own dedicated block space that is not the case anymore but we are there we have all of these suppliers and our own Services ready from day one so that you can create your app as a as a roll app on day one and you have bridging oracles everything included uh same day just

like on AWS where you know you have your mongod DB and and whatnot uh ready stripe maybe even and so on ready sort of uh in a sort of uh Lego break way so here's a quick video for how easy is to create your own rollup you just hit deploy give it a test name pick your stack Pick Your da layer yeah you can fig configure some options like on- ramping viat on ramping you deploy it and Bam it's ready um so yeah next week by the way we're launching base l3s so you can also build your application as a base L3 and and be connected to that ecosystem which is popping off right now you can see the Integrations you have on there like

layer zero and so on we support as well um yeah pretty easy try it out all right so essentially you probably heard this a lot all all of this is about the modular blockchain design and where before we had ethereum the Singleton monolithic execution layer and data availability layer we have with rollups separated the stack into a settlement layer ethereum which is also functioning as a data availability layer especially now after blobs Al Al and we have separated out this settlement layer where you basically store your you still deploy your ec2s there right like you have like

basically the asset layer sometimes call it decentralized asset layer but you put them into a rollup bridge and the rollup itself is the execution layer the rollup basically um um runs the state transition can I ask a question like while we're waiting yeah sure so if if um I have a talk on ux and DX uh kind of in web 3 and kind of in your words what is gelato doing to kind of Advance um I guess developer experience um like yeah it's a good question so I mean it's actually interesting so before rollups we approached this from the services point so like 3 years ago we released our relay API which allowed you to do send

gasless transactions it was uh used by account abstraction sdks like the one from safe for example and others and um that was nice like back then I already wanted developers to start paying for the users transactions uh so that we could do away with gas and users just go back to clicking buttons and especially with wallet as a service providers like um web R privy and so on uh this was really great we had many nice um demos where people would just sign in with Google they would buy an nft maybe with credit card and they wouldn't even know that there was a blockchain underneath um but uh in reality this never really took off that much because most blockchains were still too

expensive for a developer to to be able to pay for all of their users gas um so so actually what what I realized is that with rollup technology coming on the market and also being the way that ethereum decided to scale uh we can now go to the base layout and we can create this block space ourselves at gelato the the cheap block space for you where for the first time it's possible even for smaller projects to just pay for all of the users gas that's especially done not maybe with an ethereum L2 but more like a validium or optimium or whatever you want to call them so having something like AIT from DAC the the the Nova thing the an thing we have like three or four chains launching this month that all use this and they basically pay zero gas we we call them zero gas rollups so that

helps a lot like actually actually going to the back to the base layer and making sure that gas is cheap that that this stuff scales is I think the biggest unlock we had um um and you know that's that's better than any abstraction that we tried to fit on this in the past because you can you cannot abstract away the costs someone has to pay them at the end of the day so best way is to actually just reduce the cost there's a question there uh Li liquidity fragmentation yeah so there's going to be like hundreds of l2s using gelato um do you guys have any way in which you

are going to be addressing the fact that with each of those l2s you need bridging you need like uh you know different dexas or whatever and maybe there's some stuff you have in the works to help kind of not make things super fragmented yeah so I mean liquidity fragmentation is one of the bare cases for the world of you know hundreds of thousands of rollups but there's a lot of technology that can fix that I would ultimately I would say every entrepreneur out there this is the same in web 2 and on you still have an have a duty to attract as much liquidity as you can right like you that's just a homework that you have to do you have to build a brand you have to attract the liquidity you have to and nowadays

people do points or whatnot right to bootstrap systems that's a reality like that you cannot run away from and I believe that there is going to be some power law where I don't know the top 10 rollups maybe have the the largest share of liquidity um first of all there's a bunch of use cases that don't need liquidity per se like a lot of The Gaming use cases don't really care about it some of the nft use cases don't really care about it so there is this world but even in the defa world where where liquidity is King yeah um it's really a bootstrapping question so if you are big you're going to get liquidity right if if binance if the next binance is going to be a rollup system then they will attract the liquidity for those that are trying to bootstrap um it is nice to be able to

tap into Unis Swap and S right you lose that benefit when you're on your own rollup but we are working with espresso for example on shared sequencing technology awesome we're back by the way um where actually uh there is very in research at least very exciting Primitives now like base sequencing and so on that we're also working on where um it is feasible at least being theorized about that you can have a rollup that synchronously Taps into the liquidity of L1 or another L2 so maybe you can launch an application on the rollup and you can have synchronous cross-chain calls and atomic execution where it is possible to process trans on your rollup that Source liquidity from elsewhere um and that's the like the the Holy Grail synchronous cross chain

compos composability but even without it I think having a very nice like ZK technology that's entering the market like pulling on cdk is already sort of leading the way there with eg layer um we running that as well and uh they are like having super fast asynchronous calls like like the web 2 did basically I think that's also going to work for a lot of the use cases it doesn't support flash loans and so on but it's not worth throwing all of this away just to support Flash loans I think they're cool but you know they're not that important at the end of the day so yeah anyways um back to the presentation thanks a lot for fixing that so yeah the core blockchain design was is now all about modularization and rollups are running basically the execution layer we have the assets on ethereum the

settlement layer and we have a DA layer could be ethereum or could be Celestia I da Avail and that's how we split up the call blockchain stack um yeah it works if I point it there cool um so now beyond that core blockchain stack we now need to add a layer of the essential infrastructure so that those are rpcs right like we run our own but we also work with RPC Specialists like drpc for example um and you also need a native Bridge UI the the rollups actually comes with that um so yeah so that at the very least you can Bridge E natively from ethereum to the rollup or from arbitrum to the L3 and you need a block Explorer

so and now we're really adding this uh layer that sets gelato apart from every other rollup as a service provider out there which is okay you also need a bunch of other essential web 3D Services um beyond the basic ones to really build great applications so for example we have functions if you know AWS this is AWS Lambda basically or if you are with gcp I think their thing is called Cloud run or something so yeah you can use our serverless functions and this is really great uh not just because you know with serverless lets you basically build anything you connect to any resource on the internet your databases or whatnot and and then publish uh database on that on chain into your application but especially because modern oracles like uh like pith and and Redstone and Deon and others they all basically um

combining our functions with their data to be able from day one to publish their data on chain so now you can build a perex as a rollup launch it and same day H hook into P or Redstone data and have it be pushed on chain and you don't need to pay these huge Oracle G gas cost anymore um uh because literally on on your block space if you go for like arbitrum any trust or I don't know something else like Celestia I with any of the stacks gas is so cheap that you can feasy pay for every Oracle update um yeah relay is great as well you can add your dep UI on there you can have uh relay together with I guess one of the econ section sdks uh safe SDK for example um and so on and and you can

build great um applications now on your rollup that actually you know let the user sign in with Google and click buttons and they don't even have to know that there's a blockchain anywhere and and we also work with indexers that are of course important the graph Gold Sky and so on and and yeah there here you you see that what's really crucial as well are like cross chain messaging services like um layer zero Conex there's many of them that we work with and they let you uh also not just move assets from ethereum to your rollup but also from other chains so if your users are coming from Bas or coming from optimism or what what not you can all move them into to your rollup via one of these services and this is also a part that's quite interesting because these

asynchronous messaging services also let you connect to other chains right um so this is also like the example I gave earlier that in in web 2 you you can have multiple instances that talk to each other asynchronous this is also like technically you you don't just have to connect to other chains here you can also connect to your own instances if one rollup instance isn't enough to process all of your transactions like for example let's say You're Building World of Warcraft some type of World of Warcraft on chain if you ever played one of these massively uh you multiplayer online games um you know that maybe some of your friends were playing on a different server right like so similar here you can have a huge game you can have 10 servers across like maybe one is the US server one is the South American

server one is European Asian server and you players can play on these rollups and yeah maybe if one player needs to interact with another player at at the at the worst case they would just have to transfer the assets via one of these uh Crush messaging protocols to that other server and continue playing there so so yeah that's the horizontal scaling what I just described basically um in the asynchronous way and there the Holy Grail really is back to your question about liquidity fragmentation is have this be in a non- asynchronous way in a in a synchronous way and that's where shared sequencing comes in and there the idea is that you can run multi- instance rollups like three instances of op for example but your smart contracts it's basically like sharding your smart contracts can still

call each other cross instance synchronously um so so yeah this is like uh you know one of the current most exciting research areas you probably heard of projects like espresso base sequencing Justin Drake is talking about it a lot recently and and yeah that that would be really great because then you know uh you can essentially run a uh horizontally scalable maybe even parallelizable instance where transactions are processed in parallel but they all modify the same state uh and yeah maybe that's another another way to to scale rollup Stacks like op and Arbitron and polygon uh above the current limits I don't know where they are exactly I think it's something like a 500 or th000 TPS to to like Beyond 10,000 TPS and we're heavily researching that and seeing uh whether we can get

that done and I'm quite confident that by the end of this year we will probably have the first proof of concept on mayet and by summer already first test net so yeah ethereum is therefore not one huge World computer just like web 2 it's a web three of interconnected rollups and you can build your own rollups I really encourage you to launch your next application as a rollup not as a smart contract anymore uh to to Really benefit from these scaling uh benefits from you know having your own dedicated block space um yeah there's many other nice things that come with being your rollup there's a lot of a lot of difficulties a lot of heavy lifting but gelato does most of the heavy lifting

for you we here to help we have most of the infrastructure already figured out and uh yeah for you it's more like you know click of a button type scenario and then you can just build your application and and have your own block space and still the beauty of rollups is you're still connected to ethereum with a trust minimized Bridge that's the whole purpose of rollups with optimism you have fraud proofs right with ZK rollups you have ZK proofs that allow you to to connect to ethereum and move assets back and forth in a in a very trust Min minimized way and even if the rollup operator is some somewh faulty or fraudulent allow you to to exit the tvl or the assets on the roller back to ethereum so that you don't really have to trust the operator much at least not for for your asset

safety uh so yeah uh and inviting you to come build on gelato and um oh yeah that's I think that's a special slide that that's new so we also have node sale as a service um that's actually very interesting if you're launching a token many of the games for example do uh nowadays what we're seeing as a big Trend this year is uh if you launch a token on your own L2 um you know uh people were talking a lot recently about icos whether they should come back uh that was like the 2007 17 era era nowadays most most token launchers are airdrops or similar but yeah we we see like a new way for for token sales uh to to make it back in a more uh compliant way um making really sure that the token

is a utility and having you on L2 is a really great way of achieving that by uh for example having an ecosystem of nodes uh that need to perform certain services in on that L2 for example with optimistic rollups you might want to have verifier nodes watchtowers that actually sync the chain and make sure that there's no fraud on there that report frauds that post challenges if fraud is detected or faults are detected so yeah if you're launching your own L2 and you want to be a bit more decentralized maybe it's a defi protocol maybe it's a game or whatnot um we have node sale as a service where we give you these verifier nodes and other nodes we can even build nodes with you based on what you want to do with them and um the way the token sale then works is that

you can the community of investors that want to participate in it they would buy a license to run these nodes and then they canot mine the token right so you sell these licenses instead of the token and they would then you know run these nodes and and prove that they run them and and mine the token over a period of two two years or whatnot so um this was actually pioneered by something called Z they're on arbitrum L3 and it's really cool um I think they a bit did a big raise there and and we see many like five or 10 projects or many more doing this now and and yeah if you're also interested in that we can help you with that yeah so last but not least uh that's our ecosystem map lots of projects on here definitely not all of them but yeah for example we already uh

obviously work a lot with optimism they use our automation as well and and polygon and and arbitrum as well we have ASA ZK evm live that's the first production polygon cdk AER chain um yeah I mean all most of these projects use our Automation Services seress servic uh servic relay or whatnot like beefy year you know safe we work with a lot uh conx Stargate synthetics all of these these guys so you in great hands if you want to if you want to build with us I invite you to join our Discord follow us on Twitter gelato Network hit me up I'm going to be here all night my flight is leaving 1:00 a.m. so I'm going to be here for the after party and so on happy to talk to you guys and yeah thanks a lot for coming to the to my

talk absolutely thank you so [Applause] much should I answer a question or should we end it here we can do more Q&A if you want to do we have our next speaker ready to go let's do Q&A all right so do we have any other questions for Lewis roll up as a service here let me get you a mic could you please bring back the last slide I can't from here you want to go back a slide I can you put up the last Slide the QR code the QR code don't worry it's also on our Twitter by the way gelato Network do you

have any follow-up questions no here we go one question there hey yeah um so I think towards the end you mentioned the node sale as a service I don't one of the second to last slides curious how that works um are you allowing other allowing um partner teams to Define the infrastructure definitions to be able to run these uh node uh the node infrastructure you guys actually just doing the research to figure out how to do it yourselves and then yeah kind of Hosting from there so we are mostly backend and infrastructure Specialists so we provide the noes uh like for example if you want to run one of these optimistic uh verifier notes we have them ready for you um and uh if you're user base or investor base is a bit more

technical they can just run it as a Docker file themselves but we also have a a sort of nodes as a service where like in a serverless environment we're actually leveraging our own web play functions for this you can just like sort of have an clickable interface to deploy this node run it connect to an RPC and so on and then start you know mining the token for the front end side we're working with impossible finance and they've sort of pioneered this they have wide lbel Solutions there for the actual conduct of the sale I guess um so yeah um and we will have the first ones uh of these not sales going live next month so it be exciting yeah cool thanks any other questions raise your hand I think that's it we already had some questions I think rollup as a

service is an interesting concept uh I think most people who want to would start a want to start a rollup this kind of opens up the opportunity for anyone to do that uh are there any drawbacks to doing it on gelato um and why as a Dap developer I guess and you kind of already covered it but uh what's what's the benefit of me uh creating a new uh application to do my own rollup so drawbacks I don't think so because it's really about economies of scale um so yeah you can run your own rollup but we run what already today 50 or so of them and um we are actively investing in that and that just allows

us to bring the the cost down typical Cloud economy of scale now um there's some trust involved right you know but uh there's ways to mitigate this we are working on more decentralized sequencing technology as well where you can run your own fallback sequencers so if ours is down for example you can spin up your own one uh so you can fall back to your own servers if you want to um that's not quite ready yet but we'll be ready soon but yeah that's that's pretty much it other than that I think it's just the more scalable model to use this as a service and and yeah and the second question was like benefits I think of it but yeah there's another question there yeah uh just want to go back to Shared

sequencing um how much harder does it get to do Shar sequencing as the VM space starts fragmenting so can you run shed sequencing between say monads VM and Eclipse like I I can imagine if you have evm equivalent an easier problem but like I wonder what your thoughts are on like how that problem space explodes when these guys are all doing crazy stuff it's a good question um so mean take this with a huge grain of salt because this stuff is just in research um I I did an episode with the founder of espresso last week it's on on the on gelato podcast um on Spotify he we go really deep into that question one and a half hours uh angelato on podcast on

Spotify uh but yeah basically I think there is a way for shared sequencing to be multiv VM uh cuz shared sequencing layers like espresso are basically just like the a layers you can think of them as like similar to Celestia they just order transactions uh and this is just like Blobs of ordered data right so it doesn't really know what a VM is per se and on it's just like a its own sort of blockchain um that orders data it's like this is hyper modularization so at this point we separate out the consensus from like yeah uh and and technically you can you just need an execution layer on top to process this data and this execution layer is the VM basically right like um so so technically shed

sequencing itself is just ordering the data and this just data right so it's very generic very um general purpose and uh yeah so this blockchain can host many different types of data and then you just need a execution engine to be able to process all of it and that's where PBS comes in so ultimately I yes I think it will work you just need again like with PBS you need very sophisticated beefy Builder entities that know how to do that that know how to run these different VMS that can do that and uh that can also generate proofs for that and so on that's like the where the rollup as a service comes in again yeah so ultimately there's going to be a form of centralization to support of this but just like with PBS the idea is to offset that by having a Shar sequencing is

maybe the wrong word it should be called shared proposal I think I guess where where then you know you have a lightweight notes that can you know at least um propose uh the blocks um that are built based on this so yeah some form of PBS and so on should solve this I think Sorry by making trade-offs right there's no free lunch but yeah all right cool all right thank you so much Lewis thank you [Applause]