Eth2 Phase 1: Shard Data Chain Introduction

Transcript

[Music] hello everyone this is xiaowei from center foundation research team i'm going to give an introduction of these two phase one phase one is the middle phase between phase zero and phase two it's an important middle link to allow

us to separate the data layer and the state extrusion layers so this kind of obstructions allows the different research topic to be perceived in parallel this one will also pave the way that we can merge each one into each two world before phase two be done phase one provides several features to make the chatrons work and secure today i'd focus on the short-term and cross-thinking and cast again

a bit more than the other topics this is the data structure of the site shot block it's much simpler than the is one block or the beacon block we have the pointers that point to the short parent's route and the beacon parent roots and the most data overhead is the black body the body consists of the binary data and it doesn't have any in protocol transaction logic so that's why it is called the short data

data layer and we expect that the layer two row ops applications will be benefits from it even if there is no transaction formats because the the guest price in shot is much lower than is one so about the guest price we plan to introduce eip159 style approach for the free market the this price adjustment family is similar to the eip if the shop block size is larger then the guest price will be increased

more and vice versa there are two parts of the total transaction free the first part is the best three that will be burned on chin and the second part is the tip it's the ethers amount that goes to the short block proposers so in phase one we expect that the user will pay the free throw maybe so there are two solutions or that is to contract and in phase two the

transaction free could be paid directly through the these two execution environments so this eip could fix some important issues of the current free market economics it makes most transactions are at the same free rate and so in the normal case for the users they only need to decide if they want to pay or not it can help mitigate the situation that miner might manipulate the free

to extract more fees from users it's also a more efficient gas estimation than the current phase the current uh first prize auctions in east one i mentioned that the shop block has pointers to point to the shop parent block and the beacon parent block routes so it easy for the shot trim to reference to the beacon chain state and to finalize the shot him we'd need the beacon chance

to reference to the shot chain as well so um in the phase one beacon chain state there's a field that points to the latest sharp black route so this field will only be updated when there are more than two-thirds of validators of uh committee of a shot committee that attested the hate uh shot block is valid so with majority support we can update the beacon's trend state and this process is called

cross thinking so if a beacon block is finalized then the shot block that the beacon state referenced is also be finalized so and if the shot block is invalid and the beacon chain somehow included it then the beacon chain is also considered as invalid it's possible that the community doesn't follow

crosstalk successfully at the slot in this figure the block 101 doesn't crosstain the sha block 100 successfully so this is the orange line is the failed cross link so in this case the short state reference in the begin state will remain the same in after proceed become block 101 and when beacon block 102 from the success

for crosslink it will include both the state transitions information of the short block 100 and 101 so it can cross link both of them together in one slot notice that we almost crossed link every at every slot this kind of per slot cross link means that we have a frequent on-chain communication period it's good for having sharp block get finalized earlier

because if the beacon block is finalized the shop block that it included it also gets finalized um so it's also good for providing better user experience of crosstalk transitions because if we want to send transaction from shot 1 to shot 2 there's no direct tuner between this shot so we have to send transaction to i mean we will have to uh send the meta to

the beacon block and let beacon block to help you to pass the meta to the another shotchain the custard again is for fixing the case that if the validators are honest but since they might charge each other too much so they don't fully download uh store and verify the shop blocks entirely a challenger

can send the challenge on chen to testify if the validator has a piece of the sharp black data so because we can set each piece of the sharp block data as a chunk and use the binary merkle tree to store the data so in this figure each leaf is a chunk of the sharp black body so in this way we can mercolize the shot block body to a merkle root the validator who gets challenged

will have to respond with the miracle proof uh to the data trunk that the challenge is assigned to so if this proof is valid then okay everything is fine but if the response failed to respond in time correctly then the data will get slashed to know understand more about the cast again this is the castigan speak this is a simplified abstract

sequence diagram that shows you the basic message floor so both the challenge and the response are the beacon chain operations on chain to make it work the sharp block proposer has to generate an ephemeral costly gain key such that the gender beats bit result is exactly zero the proposal has to reveal their key before the deadline but also

it can't be too early so after the proposal reviewed their key during this certain period if other validator uses the list key to recompute the legendary bit result and if it's not zero and it's one then they can be the whistleblower to report that the block proposer has made invalid block and the black proposer will get stashed and the whistleblower will get a reward in phase zero the beacon community

members have to attest and propose the beacon block and in phase one there's a new responsibility for the big validators is to help the like clients to verify the block properly so besides the beacon community we will have another type of committees called the like client committee they are responsible for sending the aggregateable like client votes to the subnets and they will have a longer turn than the beacon community so they can

be a more stable source in the p2p network as for the short networking the main challenge is how to deal with the new shot data transfer overhead in the phase 0 current gossip subnet and we expect to do more prototyping with clients of the phase one networking in phase one at least we can expect to

get quadratic computation improvements plus with some little to roll up solutions we can get multiplied results here are some helpful reference documents that can help you to understand phase one deeply you can you can follow me on twitter or on the eastern discord server there's a phase one channel to ask some phase one related questions and we can discuss there

thank you for your listening