Anthurine Xiang, EthStorage: A Modular Storage Layer for Ethereum on EIP-4844 | ETHSofia 2024

[Music] hey guys um it's my first time in Bulgaria and really happy to be here yeah so today I'm going to introduce a modu storage layer for Ean on EIP 4844 and I'm the co-founder for E storage and cork chain so before I go for my slides um let's what a short video from C so he's my colleague the founder of uh e storage unfortunately he cannot be here I'm the only one here but he also um recall something for the is sopia hello everyone I is Sophia my name is CH Jo I'm founder of is storage web and clock chain a quick intro for myself I have been actively working in the Ean ecosystem since 2018 before that I was a software engineer at Facebook Google uh emcell probably working on their infrastructure such as high performance storage systems my background has made me to particularly interesting in E lay 2 Technologies and also data availabilities uh our team has received multiple grants from Ean foundation and optimism because our work in this area uh including data availability sampling research the development of is storage and uh the fast finality for prooof algorism so what is is storage is storage is the etherum a modular L to storage solutions that address etherum story challenges so unlike popular L Toes that aims to increase the efference transaction per second our goal is to scale air storage dramatically by up to 1,000x capacity uh I mean to achieve about a couple of PAB pedabytes and this is crucial for for future ethereum road map especially after a couple of the ethereum minute upgrade with data availability Technologies um the beum is able to accept and process tens of terabytes of data per year and without a solution like is storage uh etherum has to discard this data within 18 days by the protocol because of the limitation especially in terms of storage of the E nodes and similar to other L TOS each storage store this data off chain this large amount of data off chain while prove that there's a large number of of nodes or rabit cards that storing this data and periodically submit this proof to onchain uh smart contract our storage proof smart contracts to for verification and thanks to our proof storage algorithms we are able to enable a permissionless storage Network anyone as long as they can prove that they store a copy of this data and be able to submit this proof on chain then they're able to receive corresponding storage fee as incentive additional benefits of is storage including minimize the trust of third party data Bridges such as the data from filecoin or Obi and in addition the data on is storage is programable by smart contract and this is achieved by offering a key to large value interfaces to the applications like smart Contra applications and lastly it is modular which means that it does not require any P upgrades to the ethereum and together with all other technology advancements on top ether such as fast Final on l two and Max URL which turns AUM into a Unstoppable decentralized HTTP server we can enable a lot of new applications on top of etherum um such as fully on chain interactive website and in this way we are able to migrate most of web to applications into fully onchain such as fully decentralized uh social network or gaming I can also enable fully onchain AI so that the AI models and the promotes and coresponding outputs in terms of text or images whatever now can be fully onchain so we are now in the testnet phase and we just launched a large testnet campaign in early this year the current currently we are auditing our contract and with all this being ready we are able to launch our mainnet early next year if you're interested in EA storage if you have more questions feel free to reach out to us thank you very much and have a good day um all right so my colleague just goes through um a briefly about like what's the storage and what we can do and I will explain a little bit like details so first um so she already introduced and also for me so I have been this space for six years um have working with she for six years for the blockchain as well before that two years in finance in New York and six years in Silicon Valley for product and data science um and also so one of the initiator for uh the big Tech group group uh web3 bu. Tech um so for the team um like team mentioned we received two grants from isan foundation and three Grant from op because we have been deep in the um you know da research and also proof of storage research for the E storage and also we did like two DK FR proof research as well so what is e storage um like T mentioned before so it's a modu layer to that means like we can be part of the e equos system and also we can be layer three for any Layer Two as well so we're not like a layer one not like the ipfs we like our vve so we are part of the eer N native ecosystem like we said so we are programmable decentralized iseral storage layer too and second um so we are trusted minimize modu long-term da so we have been listened to like the eeran da theas and agen da for a while I will explain a little bit more why we are like a long-term D solution the third one is we en label fully onchain applications through web stre protocol she just mention a little bit the decentralized HTTP I'm going to explain more as well what's the motivation of we are doing the e starage um so early this year there's a dunin upgrade which is one of the biggest you know upgrade we know after that we got a new eeran DA layer which is EIP 4844 and actually that's not the end so there will be D Quin upgrade there will be D Shing so finally we're going to move from 4844 into the you know the final 4da solution as time goes by as you can imagine there will be tons of the data stored on Chen and because of that there will cause Ean too much burden to store all this data so by definition by Design the EIP 4844 is going to through the data in 18 days that's actually the first time in the history for Ean to through the data by thems because we all know like okay blockchain is going to store the data forever everything is traceable right but no after 444 that change what about other da solution for sashia for AG da actually that applies for cesia they're going to store the data for about 1 to 3 months and after that the data is disappear a year is about like 14 days then people is going to ask why we need to store the data why we care that I think the first reason which is the most important thing is all the ler TOS they need additional trust to drive the latest States if the data is not going to stay forever and the second one is if I don't tell you so how many of you guys actually know like we're not going to stay the data forever the data in 4844 is going to disappear in 18 days maybe you know few of you guys notice that right yeah so that's very surprising um actually there has been a lot of discussion for that to call for aign decentralized solution to store that that's why people are saying like da data availability actually equal to data publishing they solve the data publishing issue but they don't solve the data storage issue so that's how we do we actually work with all the da solution is4 a44 solha again da Avail to provide a long-term da solution what does that mean so you store the you know your data to any da is Da or cesia or AG da after a certain time period we're going to continue to store the data for you forever so the features for E storage like I mentioned before the first one is is we're not you know um so we're there's no data bridge for us um so we store the blobs by calling the contract directly we're are native Layer Two for the ethereum we can be native layer three for other Layer Two as well so there's minimum trust the second one is um so all the onchain storage proof verification is through the snark so we fully integrated with EIP 444 after other decorations so we will fully integrate with other you know upgrade as well and the second one which is very important we know store on etherum is very very expensive how about like other layer too like op arbitr or something they solve the issue a little bit but they only you know scale for TPS you know computation power they're not scale for onchan storage that means store on other layer to is still very expensive so we can help reduce the store cost by a thousand times compared with store on is directly 100 times compar with store store on other Layer Two directly and also we can expand the store um storage capacity to the PB level so this is an innovation we did and we received um grants from um the etherum foundation so people ask me if I pay one times how can I guarantee my data store forever so even we charge you one time we're not going to give the storage provider one time fee so we give them the fee as time goes by we need to make sure as time goes by they store everything they can go the rewards but for you as the user side you only pay one time um yeah so this very simple about like how does a structure looks like as you can see so each storage is ler to network there's a lot of storage provider um one thing I want to mention here is you don't need to buy any machines as long as you have like eight taby storage capacity you can join as a note yeah so the entry barrier is quite low and on the etherum layer ones there's a smart contract every time so when something stored on the layer one so the contract in the layer one we inform all the provider in Layer Two see the data is coming you need to download that so everyone is going to downloads into their story provider after they download successfully they have to provide approv of storage to prove they really did store everything and a smart contract in L1 so they will verify it's true or not if everything is true they will return the rewards which is EAS um yeah so we have been working for this for two years as she mentioned for this year we did a lot of things we integrated with 444 and the P testnet has been running for about like six months so we integrate with sashia and other Solutions so we're working with OP stack the cdk integration and other layer to like Tao M as well so the progress like I mentioned the public test n is live we have conduct two rounds of campaigns so far um there ends unfortunately but there going to be other campaign soon and this is announced early this year in E Denver so um we got like um you know public uh partnership and integration with cesia and this is um we receive the grants from other layer to Tao and op like I mentioned before although we are all there too so we're actually working very closely because they're all optimized for a TPS computation but we are optimized for onchain storage so we can help them to achieve more in combinations you know so in that way so there will be more um DF and one common question I got is what's the difference between you guys with fire coin or arve so there three of them the First Ave and fire coin they're layer one they're single blockchain they have their own ecosystem we are part of the ethereum ecosystem so we're kind of like the native five coin or native arve for the whole e ecosystem so that's the first one the second one they can only store static data while us we can store Dynamic data in that ways that means like we can fully programmable see here the third one onch programmable yes we can do that and the last one because we are fully compatible to the E EOS system so anything you already use for example like a metamask and other tooling you already get used to that you can use directly you don't need to learn another um you know like a stack or coding skill or whatever anything you already familiar with just use that one thing we we just mention a little bit remember what's a third feature so we can achieve fully decentralization onch D app through web3 protocol web3 protocol is another product we build working with e storage very closely it's a decentralized HTTP yeah it's very easy to understand so currently while order the app they claim their decentralization but we know the website they're hostings they centralized right there's DNS there's infura and a lot of attack for defy actually came from the DNS Tech because they're centralized component and if there's censorship Shing like you have to shut down your website they have to do that through the web3 you know protocol decentralized HTTP we put all the components in the fully decentralized way so that means like there's no DNS because we use ens or other decentral name system you can choose whatever you like and also there's no infura because it's hosted in the fully decentralized way because of that so there will be a lot of new application came out the decentralized social network and also fully onchain AI um remember so currently all the decentral AI how does that work so they run the model offline and then they put the data or the results on CH why does that because there's no space for them to fully run everything on chain through e storage so we can run a large AI model onchan make that possible and also the onchan interface the promote into the AI model into the output um so we're working that with aura as well yeah other things like the fully onchain games also has a limitation of capacity that's possible enable for us as well yeah that's pretty much everything um I'm talking today happy to take any questions and also I will stay here for three more days yeah just grab me I will always be in the conference venue thank you guys can you guys hear me yep thank you for the great talk I was wondering about the web tree protocol the URL scheme that you also proposing in addition to the EAP uh has there been any program with browser vendors on adopting this uh uril scheme um Sor I don't I don't think I hear you so you're talking about which EIP the I'm talking about the web three uh Euro scheme I was wondering if there are any browsers that uh adopting this seral scheme yeah yeah um you know for web protocol so if any big brows for example like the Chrome or like five Fox can support that that would be awesome right but that's definitely some big progress like we all of us need to working on that so right now um how you access that for this big brother you have to use extension we do have a native um crypto BR that's our community member helped to develop that so that BR is native support that that's the only one for now but I hope in the future there will be more brother support that natively yeah hey um how's it going thank you for your fascinating work um I have a question about um the structured data in the way it's stored so in a lot of web applic we expect some sort of structured data indices and other metadata that is kept along the real data and that what that's what makes the web pretty quick today as such have you guys explored um any ways to perhaps merge sort of the data block storage with the highed order structured data index data perhaps uh relational database of sort or a nosql database what is your thought on this thank you okay so so let me try to understand that so you're asking about like what kind of data we store or no um when you when you write data to a hard drive for example you break it down into blocks right but to a SQL database let's say a post gray that's not very useful as such right he has to load it up parse it create indices on the data so that the lookups are quicker right it's a data access question really I'm not sure whether I hear you clearly but we can offline discussing more um so for is storage you know so first if you think you have like a you have some issue because store on the centralized way it's very expensive store on decentralized way might not be more cheaper so that's the first one so if you only care about like price that's not how you choose between like centralized or decentralized the second one is for fire coin where Ave other store solution we're not cheaper than them so you're if you're you know combined between the different like decentralized solution we're not doing that but what we can do more is first so because they can only do static data so we can do Dynamic data that's the first one if you are choosing about like Dynamic data we will be um you know much better for that and the second one is if you are within the etheral ecosystem you don't need to store the data in other chain and do cross-chain transaction so store on us we're native for the whole eer equal system so that's the second choice but if you say like you are sonana or s or other later one blockchains that's totally different story because even store for o or ipfs or Ave they're all cross chain solution CU they're all independent I'm not sure whether I answer your question uh it was more around like I'll give a very simple use case imagine that you are here in Sofia and you have a bunch of objects around us let's say coffee shops and you want to sort them by some sort of a gist data like distance walking distance in a straight line let's say say that requires structured data and and some very interesting data manipulation so that you can actually have a query response within a reasonable amount amount of time right so that requires some sort of a higher order use of the just blobs stored on a some storage somewhere right so I was I was thinking whether I I wanted to ask whether you guys have spent some thought process on on this specific issue yeah I think we can discuss more offline yeah so just live time for other one thanks [Music] I'm I'm going to