Solving Vitalik's trilemma with zk-driven DA and Storage

Transcript

[Music] thank you guys for coming for the next speech and uh we will welcome Iger gulamov and he's going to so uh to give a talk about solving vitalic trma with the ZK driven da and storage welcome um hi so I will tell you about our U solution uh of metalic trma with ZK driving dat availability and

storage so what is vitalic trma we have three property oh it looks like technical issues happens okay now it's better so uh okay what just Happ it's blinking uh uh what does mean with Al um uh it's it's not not corresponding what I'm

showing on the screen this one okay okay um so uh what does mean with Alex trma uh we have uh three properties um of decentralized Ledger it's scalability security and decentralization uh and uh the trma means that we can select only two of three uh why uh the solution of vitalic trma is important uh because uh we need foras adoption scalability um especially for uh migrating from web 2 to web 3 you know

Solutions like foraster which need a lot of storage transactions and so on and with very low costs uh so uh that's why we need uh uh the solution of italic stremma to make possible another more heavy Solutions not only decentralized Twitter but the centralized Instagram Facebook and maybe YouTube uh uh the historical review of this uh uh solution um is uh uh the first it was multi chain EOS system proposed by poka do and Cosmos the idea is that there is

core engine uh on which multiple blockchains could be implemented and uh this solutions can use this core engine uh for interchain uh Communications uh next it was plasma uh based on uh Game Theory and if something is going grown anybody can start uh the game uh in which he can wait itat and get his assets uh back uh uh next uh it it was uh evolution of plasma it is optimistic rollup uh the K

rollups and the K validium and the most fresh solution is data availability layer it's implemented in ethereum uh in Protank charging and also the there are solutions like Celestia agent layer and aail uh who provide data availability also uh the issue of data availability is that uh the uh data is stored for short time it could be 2 weeks or 1 months um and uh this uh things are originated from plasma when um it is enough uh to absorve the operator during 2 weeks and

uh then if you have some assets or something else uh uh you can save the data you needed and then you can uh make transactions with this data data uh uh why rollups are not enough for solving the vitalic stma uh because rollups uh don't scale the data so here is uh how rollup is working uh we have uh uh L2 with State transition proofs uh but also we have the data and we should publish the data with high soundness way because

all users need this data to make transactions uh users cannot make transactions uh uh for all up if they uh have no data uh also that's why uh we cannot uh deploy rollups into rollups and so on because the data is floating up the data is floting from L2 to L1 uh before darding it was stored in cold data uh now it is using dun sharding and it also could be used another data availability Solutions uh but anyway this is

bottleneck uh and also current data availability Solutions are storing the data just small time frame and the next uh the users should uh store the data bys or they should trust uh any of chain infrastructure uh uh sort parties who do it we need recursive rollups to solve vitalic trma uh and we need to solve this issue of floting up the data uh I our approach uh is data Centric

Solutions um the the current Solutions are file coin uh rwa and a storage uh they uh use uh data availability sampl uh for uh proof that the data is stored uh and available uh and uh um uh also uh miners can you uh can mine the rewards using the data stored by nodes uh uh to solve the vitalic trma we need some improvements uh the first

first is uh using read slamon codes for more than 10 times better uh um storage utilization because current Solutions are using replication and uh what does it mean that mean that if we we need High soundness storage for the data we need to make a lot of copies of this data and hope that uh at least uh uh one of this copy will be alive and then uh the the data is stored and available um uh with r amount codes we can mix all the data together and make blow up and

then if any uh Source amount of uh new blown up data is available then we can reconstruct the source data uh it is uh uh much more effective efficient solution than just replication and we can reach more soundness on it uh and U the next one thing is wall chain uh provable um Z proof uh it's possible it's shown by Mina blockchain that uh we can do a lot of uh State transitions consensus related things and uh so on

and we can compress this all into one single proof uh in our research uh we shown that all uh uh uh data specific things uh we needed for uh uh blockchain fun uh data Centric blockchain functionality uh it's possible to prove uh with zik snarks in one single proof and it is working like uh um ZK data availability and storage Oracle and this is very useful thing if we need uh uh to solve itma and build uh recursive rollup

system uh speaking about rlon codes and its efficience uh here is how it's working um uh let we have four numbers we can encode it as uh a 3 degree polinomial like this then we can make blow up we find another points then for example if we forget uh some part of this data now we have another points but any four points are enough for fix it uh uh degree 3 polinomial and that's why we can recover the source data and uh this uh uh is much more efficient than just

to take these four points and make a lot of copies of it and hope that each point is present at least one time because if the scope of data is uh large um uh there will be high probability that at least one point will be lost and that's why re readon codes are better because we uh can prove that uh uh all data is safe uh speaking about Ron Cod the important problem is how to prove to all data storage uh not that their shards are

correct and we can do it with polinomial commitments so we have polinomial commitment to Source data and we can make this polinomial equation to prove that commitment uh of the shart for each node is uh correct readon Shard uh that's why all uh uh nodes could be ensure that they receive the the right chunk of data and uh that's why uh any anybody can collect sum of shards uh any subset uh of these shards we use blow up factor 8 uh it is 64 to 512 so so if uh

anybody collect 64 shards from uh 512 shards uh uh they can reconstruct the source data uh here uh how uh data Centric Roll Up is working um we have uh data storage Network L2 L1 uh um so idea is that uh we can make zero knowledge proofs that the data is stored and available uh we can do it because we can prove with narks all process related to uh storage and availability of data

including State Transitions and um random U data availability samplings uh and uh um mining of nodes um and so on so we can make zero knowledge proof that the data is stor available and we can join it with rollup execution proof and by this way we can uh construct the proof uh with uh uh State transition and data availability inside and when we publish this proof on chain we don't need um additional things uh um that uh um like floting up the

blocks because this Pro with this proof we already know that uh the data is stored and available and uh users can use it uh so here are some parameters of such rollup the capacity is more than one PAB the soundness is more than 110 bits that means that if half of network is malicious then uh uh the data uh could be lost with probability uh to uh in order minus uh

110 very low number and uh the cost if the node is uh hner node uh is 15 cents per gigabyte per year it's including eight time blow up already uh in specialized rigs for uh storage the cost would be lower and uh recursion is unlocked because this rollup uh provide data for itself that mean that uh such rollup could be deployed inside rollup and uh we don't need to uh flood up any blocks and we can build uh decentralized infrastructure of rollups

uh next could we reach more scalability the answer is yes uh because uh uh we unlocked recursive rollups we can build a network uh of rollups providing infinite scalable system uh only one overhead here is that we need uh to manage uh mining nodes on top layer to prevent concentration of Mal malicious nodes in one single segment of the network we have modeled uh all this uh um situation and we have strict proof that this uh uh soundness estimation is valid for asymptotic infinite scalable

system with such architecture uh so uh how it could be reached uh when when somebody uh left uh from the pool uh we should find to the replacement node and also we make two random mixings and this is enough to prevent concentration of malicious nodes in one single rollup and that's why we can build uh uh infinite scalable uh uh system like this uh so uh here are some links uh for our research uh I think uh solution or vitalic stemma is important

uh for for uh migration from web 2 to web 3 uh not only decentralized Twitter but decentralized Uber decentralized Instagram decentralized Facebook and so on and uh of course it's important for uh general purpose blockchain specific problems also thank you for your attention thank youer do we do we have questions silence no questions thank you very much thank you for your presentation and thank you for your attention uh if you're attending any side events please go and uh tomorrow we start here 10:30 10:30 see you back then

thank you [Music]