Data Availability Summit 🗄 Blobstream: Streaming DA for Ethereum

[Music] so I'll talk about uh blob stream today uh as cartic mentioned uh this was formerly called actually quantum gravity bridge but that was a terrible name so we renamed it to blob stream recently uh there's a blog post about this that's linked to later in the slides uh if you want to check it out uh so now let me talk about what blob stream is uh it's General architecture what is benefits are over other data availability scaling solutions for ethereum so first let's talk a little bit about Celestia to give a be brief background of blobs and then we'll talk about the streaming part of blob Stream So when Celestia is the first modular blockchain Network it provides a highly scalable uh data availability layer that you can build execution layers on top of or you can stream or relay uh attestations of data being available to other chains such as ethereum which is what blob train will do uh what is the benefit of uh being a data only layer uh Musta talked about this earlier in his talk right it means that you don't have the overhead of execution settlement all their stuff to deal with you can just optimize and maximize for data availability and therefore for the same cost to run a node you can have substantially higher data availability throughput than a traditional monolithic chain would have uh and of course the data availability is the most based thing it's the basest of layers uh and that's the thing that everything else can be built on top of so the more capacity you have at the base layer for the same cost the better it is because you can always build more on top of it but you can't really add more at the base layer without making sacrifices to decentralization Etc okay so uh architecturally and forgive the terrible diagrams these are made by engineers not artists uh Celestia has these things called blobs and they were these have been renamed twice already don't don't ask what they named before uh blobs work as follows uh they are uh indirectly referenced in pay for blob transactions so this is the TX the blue that you see in the Square uh and Celestia squares are laid out as squares and why they're called squares or Celestia blocks are laid out as squares and why see the squares in the picture here we see two two examples of blocks of varying sizes uh the the pay for blob transaction is in blue The Blob itself which contains the data for the execution layer or the L2 or what have you that's in yellow uh and the this blob is committed to in the transaction in other words there's a merized commitment to this blob uh and this indirection uh as opposed to directly including the blob and the transaction which which is what ethereum does with its call data this indirection allows you to to have what we call partial nodes in other words they're nodes that can fully verify the Celestia State machine without having to fully download Celestia blocks we have this new class of nodes that all their traditional monolith monolithic blockchains or ethereum specifically if you want to say uh don't have because they don't have this level of indirection it allows for a pretty significant scale while having this very secure not type that's good uh now these blobs are committed to in transactions they're also committed to in a Merkel tree as is tradition you know block data is committed to in a Merkel tree have Merkle tree then you have other Merkle trees and so on there's various layers of commitment here uh that was the background on the Celestia now let's talk about the bridge SL stream part uh and we'll fit these two together in a couple slides so traditional Bridges uh operate as a bidirectional asset Bridge so they move for example the Axel Bridge uh I think maybe there's also an IBC bridge that was out recently I forget uh move can move funds between two chains and they move funds specifically or they can move messages that kind of create actions either or uh and it's bidirectional these are traditionally how bridges operated and whenever someone says bridge this is basically what people think of uh there's pros and cons to this uh the pros of course that you can move money around the cons is that if the bridge is compromised then this can result in very significant loss of funds because the bridge contract or Bridge applications are kind of the be all end all if they're compromised then all the funds can be lost and we've seen this in a variety of bridges such as Wormhole and so on uh how does blob stream work it's different in two primary ways to these bidirectional asset Bridges which is that it's unidirectional and it does not Bridge assets over it only Bridges data or it only relays stations of data being available the zeros ones here uh so it will stream kind of like how you know Netflix streaming is not you upload to Netflix and Netflix downloads to you it's one directional right Netflix will stream data to you and that's it you see a movie and that's it uh so blob stream will stream data from Celestia to ethereum not the other way around and not assets uh this has a number of benefits uh one is that uh one was that actually I'll talk about this this security benefit after I explain how L tws would integrate blob Stream So how do you use bobst stream if you're Layer Two uh such as most recently arbitrum orbit I believe just a few minutes ago got tweeted out that arbitrum orbit uh has uh does an integration with uh Celestia for data avail a ability so we'd have some some system like this layer two uh it has to Post Its transaction data somewhere uh in the traditional rollup scheme mechanism it posts its transaction data directly to ethereum as call data very expensive just absolutely very expensive and more importantly it has a low capacity so it's expensive now but as the rollups get used more and more it'll get even more expensive because this data has to compete with everything else uh to give you an idea on order orders of magnitude if ethereum today has approximately dial up uh levels of bandwidth Celestia with its uh default block parameters block size parameters is approximately like a band a Broadband order of magnitude so we're looking from like going from dial up to broadband that's the order of magnitude capacity increase uh that we're looking at by using a system such as a bloodstream or just less in general for data availability so this opens up you know the door for things that just straight up aren't possible in ethereum today so you have L2 it posts this transaction data somewhere in this case if it's you know blob stream L2 it'll post this transaction data to Celestia the Celestia validators will meriz and batch together a bunch of Celestia blocks into a single merkler route and then relay this attestation on this batch to The Blob stream uh da contract right this will verify the attestation uh and then Downstream L2 settlement contract uh so these are you know V arbitrum for instance you know the it's the contract that verifies data availability and that adjudicates a fraud proof if your Z rollup you know it's the one that verifies the proof and also checks that the data is available th those kind of contracts can those Downstream contracts can then verify that testation from The Blob stream da contract uh with very with very little change to their current logic honestly uh so what is the uh what is the nice thing about this the nice thing is that this relay contract this this relaying of attestations is not actually the be all end all of security what is the BL end all of security is still the layer to settlement contract so even if there's some sort of compromise in uh the bloodstream contract or in the bloodstream relay mechanism the only thing that does it does not steal funds the only thing that it can do two things one it can stop relaying data which is okay it's a temporary liveness failure that can be required from the other thing that can happen is that an invalid Merkel route can be relayed over uh but as long as the L2 contract is honest and as long as those sequencers are honest uh then it doesn't really matter if if the if the if the attestation is incorrect because that attestation will just be rejected by the sequencers uh so it's not a is not a weakest link uh thing between the the the blob stream da contract and the L2 settlement contract it's a the settle the L2 settlement contract is still the BL and all of security in this gamee and this is different than you know bidirectional bridge where the bridge contract is the thing that provides all the security okay so uh what are some of the benefits now of blob stream versus data availability committees and reating type schemes uh this is uh so other than posting data to ethereum various L2 sidechain projects have suggested uh doing a variety of things for data availability data availability one of them is using data availability committees so these are a trusted small set of you large ecosystem participants or whatever does effectively Leverage The reputation to say we won't do anything bad please trust us uh and then they this committee will sign over some data being available or some data even being valid actually why not uh there's also reaking uh reaking involves taking the ethereum validator or reaking on ethereum involves taking the ethereum validator Set uh and allowing that set of validators to also attest to other things such as this data is available or this data is correct uh now what are the downsides of data availability committees and also restating which are very close to each other in terms of the security guarantees they provide despite the fact that reaking is much more complex they're actually very close to data availability committees uh what are the downsides is that if you do something that's Pur reputation based well you don't have any security right uh it's purely reputation based so we're back to the traditional Financial system uh if you have some mechanism to penalize the committee or the rakers in other words slashing uh you have to ask yourself who does the slashing and this brings us back to what Nick was saying who watches the Watchers uh and there's a few ways you can do this the most common way that's suggested is token weighted governance onchain governance uh but of course there's the issue and exactly as Nick said it's not the validators that you want to control the chain or the protocol it has to be the users the developers the uh the developers of the protocol and developers of applications on top of it uh and also token holders to a certain extent right it's the entire community that you want to have uh governance over the chain SL protocol not just the token holders and not just a subset of those token holders the validators SL people who would vote on governance uh if that's the case then as Nick was saying in his talk is it degenerates to the what we have in the traditional Financial system very much so if the only way you have to slash people is token token voting uh then that's does that does not provide cryptoeconomic security right uh because the token voters could just decide not to there's no penalty for them not to slash now uh there is another way to slash and that is with fraud proofs for example right you might say well I mean uh optimism and arbitrum have fraud proofs and they have this thing called slashing why can't you just do that so the reason that you can do slashing with fraud proofs is that fraud proofs are over an objective thing objective thing uh you know you object L and you you can also guarantee the availability of that objective thing uh based on your fraud proof scheme uh then you can slash People based on an objective fault that they have committed data withholding uh in a data availability context uh as we've known for many years is subjective it's a non attribut it's not attribut attributable to a single entity uh and in such a case you cannot objectively say that data is being withheld and therefore you cannot objectively have a fraud proof that can be slashed you must rely on either token governance which as we just said is not ideal or you can rely on offchain slashing now data availability committees and reaking do not allow for offchain slashing layer ones do allow for offchain slashing uh such as ethereum right ethereum's native proof of St protocol and the reason that it provides cryptoeconomic security and the reason reason that people trust ethereum is not just because they trust the validators it's because they trust the governance of the ethereum community will keep the validators in check and will slash them if the majority of validators are dishonest that's where ethereum really derives the security from uh similarly if you have another layer one such as Celestia the community can keep the Celestia validators in check even if a majority of them are dishonest so in in other words to do slashing to provide this cryptoeconomic security you really need it to be something that happens off chain it cannot be token weighted governance and this is where blob stream is different and strictly better than data availability committees and reaking it's because since Le is a layer one uh the community can slash the validators if they misbehave in ways that the the other two after mentioned things can on okay so we're wrapping up on time if you'd like to find out more uh there's an introductory blog post that was published a few days ago uh that reintroduces blob stream under this new name uh and that provides it additional contacts and that goes over a lot of what I discussed here we also have validator docs so if you running a if you want to run a Celestia validator and also want to uh attest to this data being available at the second link there's also developer docs so if you're building an L2 on that settles on ethereum right something like a rollup oh sorry for that was a car passing by you know something like a rollup that settles on ethereum and that would like to leverage SL data availability that's a third link and finally if You' like to look at our smart contracts Direct The Blob stream da verifier contracts as the fourth in GitHub so with that thank you for listening and I hope you check out blob stream when you get the chance