Layer 2 Stage | NEAR as a Data Availability Layer for Ethereum Rollups

Captions as YouTube publishes them, usually machine-generated, so product and people names can be misheard.

1. 0:00

   next up we have rock climbing extraordinaire fat sgz who's going to be talking about near as a data availability layer so please welcome fat here hey everyone uh my name is fat I work at pagota as an engineering manager and today I'm going to be talking about something that we've been cooking for a while um data availability layer when near so Ilia kind of explained yesterday what we're doing um

2. 0:34

   on data availability layer for near but I'm going to be doing a little bit more Deep dive and then we're going to look into how we do this for ethereum rollups um let's see if this is going to work this not working guys oh it does okay yeah so what we going to do today we're going to be looking into what data availability is uh we're going to be looking into what what challenge that we have with data availability we're going to look at what a rollup is um how many

3. 1:04

   people know what a rollup is here oh wow okay then I'm going to be quick um I'm not going to go too deep or maybe if you want to go too deep we can go there um why da is important for rups right um we're going to introduce near data availability layer we're going to do component Deep dive and then we're going to look into what's next for the for the data availability layer um what is data availability in the first place um data availability is a principle um that data that is stored

4. 1:37

   on a network is accessible and retrievable by all Network participants yeah okay um cool that is very easy to understand right you just make data available to everybody who's in the network but what is the challenge the challenge is um when you want to scale the network because the problem is that you don't want everybody to verify all the data naively um you want to have some guarantee that the data is available to the participants but not make everybody download and verify all the data all the

5. 2:10

   time um but what is how does scaling work in the blockchain space um there are two thoughts on this one of them is what we're doing at near protocol we're doing onchain scaling um but the idea is that not everybody should be verifying all the data some of them be verifying some of it some of them should be verifying the other part um sorry validating um and this is how we're trying to solve this at near protocol right we're trying to do this

6. 2:40

   by sharding which is an on chain scaling solution how ethereum is trying to do this um is kind of different again same idea some part should do some some some some part should be should be doing the other part um and it's called offchain scaling there's multiple offchain scaling types so to say um but we're going to look into rollups today right because right now that is the one that is kind of um picking up

7. 3:07

   what is a rollup um in ethereum World currently um what they're trying to do is instead of having a monolithic blockchain they trying to have a modular stack right where an execu there's an execution layer there's a settlement layer and there's a consensus on data available to layer um what they're trying to do is they trying to take the computation out right the execution out of the rest of the stack put it in a very let's say a

8. 3:37

   super node or a bunch of super nodes do the calculation roll up the transactions put them in a bundle um compress them and put it back onto an L1 which actually scales pretty good right because you you don't have to um not all network has to do all the computation but then there's a problem how do we know that the computation that has been done on the on the super node is actually correct how do you prove to the N to the

9. 4:10

   rest of the participants that your state transition is correct let's go um so right now there's two ways of doing this and some subclasses of fit um but first one is optimistic rollups right so you have to the idea is that these supernes optimistically um produce these State changes and state change commitments and then put it on an L1 and then just they keep doing it and they wait for a certain amount of time where if somebody

10. 4:49

    who also has a stake in that Network comes and says wait a minute there's a problem here's my proof which is called the fra proof and then the chain R rolls back the L2 chain rolls back that is why it's called optimistic Roll-Ups and for ZK um the proof is done on every execution and it is called a validity proof I am not going to go into detail on ZK on technology I just called it cryptographic magic um but at the end of the day every execution comes with a

11. 5:21

    cryptographic proof that the execution was correctly done how does that relate to data availability um so we said that you have to somehow trustless verify the correctness of the computation performed offchain right but how do you how do you create the fraud proof in the first place there's a state a and then you want to go to the state B there's an execution between which is done through transactions so you have to

12. 5:51

    have access to those transactions in the first place to say wait a minute there's actually a problem with these State changes and that data has to live somewhere and it cannot really live with the the the party that does the execution because what if they just don't respond what if they just go away right you want to be able to continue that chain and keep the state somewhere for the ZK you don't really need the transactions but how do where do you check your balance for example it

13. 6:23

    has to live somewhere um and that is where the data availability layer comes into play so what you could do is you can just since you're already putting your transaction State commitments onto ethereum you can just put it on ethereum which is fine and it has a high security but at what cost ethereum is currently and with without um prot dang sharding or D sharding it is quite expensive to post this data directly on ethereum so what did we do then um we

14. 6:55

    said hey near protocol is has very cheap storage has cheap um transaction processing why don't we just create a contract and then let these l2s put their data their state or their state diffs or their transactions whatever they want to put it's just a bite array at the end of the day directly on near protocol and provide them with a light client where they can say like hey this is the transaction did it actually get inside the is it actually in the chain which is

15. 7:28

    what we call inclusion proofs and then we created a really minimal RPC client for rust and non-rust interactions so right now we have it we tested with go and rust um and that is pretty much it actually so if you look at this graph you can see that there's ethereum the roll up deres blocks executes transactions and post commitments on near through near da we put it on a contract there's a light client and then the fisherman which is fishing for

16. 8:01

    frauds right to create fraud proofs um is asking the light client is this is is this sin is this sin is this in and at the end of the day can go back to eum and for the settlement and say like hey I found a I found a fraud and that is what we build the the blue parts that you see is what we've been working on for the for a couple of couple of while now couple of months and we're very excited it is live by the way right now I will give you

17. 8:30

    the the GitHub Link at the end if if anybody wants to check out feel free to check it out why would you build on near why would you build on near da layer um well it's very very cheap right it's for now we also found a optimization it's now it's 8,000 times cheaper than posting on eth um we're going to be working on it to optimize it more we but something that is really important we we already have a bootstrap

18. 9:00

    Network that is has a 100% up time for the more than a year 3 years a year 3 years now um which kind of says that it is you can trust it a the the the security of network is already running and working the validators are already up and running so the bootstrap um of the network has already been done we already created a lightweight da client that is easy to run um lowcost running and we have fast final finality own State

19. 9:31

    data so this is really exciting for us um what's going to happen next right we already released this open source you can go and check it out you can if you want you can integrate your own rollups to work with near da um IL announced on Partnerships we're working on more Partnerships coming up um from a tech standpoint what are we going to do next um we're going to be working on data availability sampling which is really

20. 9:53

    important to um to even get the cost down further and uh introduce data redundancies on near we're going to be working on we're we're working on 2D Erasure coding um for the for the data redundancies we want to improve the commitments we looking into kg commitments which is going to be getting us closer to um ethereum we want to lower the barrier to entry we want to create room to uh sorry

21. 10:26

    room to create economies around this uh um the as clients um we want to create non interactivity with um eraser code commitments you saw that there's a fisherman we just want to get rid of it it's just an additional actor we with kzg commitments we don't have to do this anymore um we want to we want to have room for more optimizations with multi proofs um sampling individual cord words and something that we want to also investigate and we're investigating with

22. 10:55

    IG layer is double Corum AVS um which means that the security will not be only on near protocol validators but also ethereum resters on these Services which which is a combined security at the end of the day and obstruction of data will be a slash offense in this um in this Paradigm where basically nodes if they lie and they'll give you the data um they will be slashed we are very excited um to launch

23. 11:27

    this new product and please go and check out um if you want to take a look at what we've done we've done already couple of um interesting Integrations um polygon cdk and op stack Bedrock we're working on multiple more and yeah I'm very very excited to share this with you and thank you