当前位置：网站首页>What is the "danksharding" of V God Kop on Valentine's day?

What is the "danksharding" of V God Kop on Valentine's day?

2022-06-29 05:58:00 【Babbitt information】

Beijing time. 2 month 14 The night of Valentine's Day , Founder of Ethereum Vitalik Buterin United Ethereum Foundation （EF） The researchers Dankrad Feist Held a conference on capacity expansion solutions “Danksharding” Educational seminars for , If you want to know how blockchain is increasing “ De centralization ” and “ Security ” Attribute and realize large-scale capacity expansion at the same time , So this seminar is a good starting point .

notes ： About “Danksharding” What it means to Ethereum , Readers are advised to read first 《 Learn about Ethereum “ Expansion killer mace ”danksharding》 This article .

The following is from Dankrad Feist Provided 《Dude, what’s the Danksharding situation‌?》PPT, If you want to watch the full seminar video , You can visit the official website of Ethereum foundation youtube channel ‌.

summary

1、 What is old ：

（1） Data fragmentation ;

（2） Use KZG Promised data availability ;

（3） The original data is partitioned using the separation and segmentation scheme ;

2、 What is new ：

（1） Proposal is - Builder （ Data generator ） Separate (PBS);

（2）crList;

（3）2D programme ;

（4） Proposed architecture

3、 Summarize the advantages and disadvantages

One 、 What is old

1、1 Data fragmentation

1、 by Rollup And other capacity expansion solutions to provide data availability （DA）;

2、 The meaning of data is defined by the application layer ;

（1） Ethereum basic consensus （ Full node and verifier ） Assume no responsibility except to ensure that the data is available ;

（2） Why is sharding not performed ？ Compared to native execution ,rollup The efficiency of the two-layer network should be high 100 times , It's hard to see what needs to be done to implement sharding ;

3、 The goal is ： Offer about 1.3 MB/s Data availability layer and complete sharding function （ Of the current maximum data capacity 10 times , Of normal capacity 200 times ）;

6、 since 2019 Since the end of , Data fragmentation has always been the goal of Ethereum ;

1、2 Data availability sampling （DA sampling）

1、 Want to know through O(1) Work earned O(n) Data is available ;

2、 Ideas ： Distribute data to n individual chunk In blocks ;

3、 Each node downloads k individual （ Randomly selected ）chunk Block ;

1、3 Erasure code （Erasure coding）

1、 Use Reed-Solomon code （ Polynomial interpolation ） Extended data ;

2、 for example , At the coding rate r=0.5 when , That means anything 50% Block of （d0 To e4） Is enough to reconstruct the entire data ;

3、 Now sampling becomes efficient （ for example , Inquire about 30 Random blocks , If all are available , exceed 50% The probability of unavailability is 2 ^ (-30)）

4、 however , We need to make sure that the coding is correct ;

1、4 KZG promise （KZG Commitments）

1、 polynomial

2、 promise C(f)

3、 assessment y = f(z)

（1） The certifier can calculate the proof π（f,z）

（2） Use C(f), π（f,z）, y as well as z, The verifier can confirm f(z) = y

4、C(f) as well as π（f,z） Is an elliptic curve element （ Every 48 byte ）

1、5 KZG Commitment as the root of data availability

1、 take “KZG root ” Imagine something like Merkle The root thing ;

2、 The difference is “KZG root ” Promised a “ polynomial ”（ All points are guaranteed to be on the same polynomial , and Merkle There is no guarantee of this ）;

1、6 Separation and fragmentation proposal

Two 、 What is new

2、1 Proposal is - Builder （ Data generator ） Separate (PBS)

1、 Invented to counter MEV Leading to the trend of centralization ;

2、MEV It means that more mature participants can extract more value than ordinary verifiers , This means advantages for large ore pools ;

3、PBS In a separate role “ contain ” This complexity / Centralization , And have a few honest assumptions ;

（1） Proposal is （Proposer）= Verifier , Honest majority assumptions imply a high degree of decentralization ;

（2） Data generator （Builder）= Separate roles , A few honest assumptions , It means that decentralization only needs to ensure an honest （ Non review ） Data generator for ;

2、2 Review resistance options – crList

1、PBS The danger of ： An efficient builder （Builder） Sure （ At a certain economic cost ） Permanently review some transactions ;

2、[ notes ： In today's bribery model , The cost of reviewing a transaction is the cost of permanently bribing all proponents without including the transaction ;]

3、crLists Allows the proposer to specify what the builder must include tx list , To restore the old balance ;

crList（“ blend PBS” Design ）

2、3 KZG 2d programme

1、 Why not KZG Code everything in the commitment ？

（1） Need a supernode （“ Builder ”） To build and rebuild in case of failure ;

（2） We want to avoid the validity of this assumption ;

2、 The goal is ： stay d individual KZG Code in the commitment m A shard blob;

（1） If we were naive enough to do this , It needs to m * k Samples , So there is too much data ;

（2） contrary , We can use... Again Reed-Solomon Coding will m Promises extended to 2* m A promise ;

2、4 KZG 2d Scheme properties

1、 All samples can be directly verified according to the commitment （ No proof of fraud ！）;

2、 A constant number of samples ensures the availability of probabilistic data ;

3、 If 75%+1 Samples of are available ：

（1） All data is available ;

（2） It can be refactored from a validator that only looks at rows and columns ;

（3） Nodes are not required to observe all the situations ;

3、 ... and 、 Put together is Danksharding

1、 Execution blocks and sharding blocks are built together ;

（1） We used to need a sub committee , Because each piece blob May not work alone ;

（2） Now? , The responsibility of the data generator is to make the execution block effective , And all the pieces blob All available ;

2、⇒ Validation can be aggregated ;

3、1 Danksharding Honest majority verification

1、 Each verifier chooses s = 2 Random rows and columns ;

2、 Only the allocated rows are proved / Whether the column can be used for the entire epoch period ;

3、 An unavailable block （<75% You can use ） You can't get more than 2^(-2s) = 1/16 The proof of ;

3、2 Danksharding restructure

1、 Each verifier should refactor any incomplete rows they encounter / Column ;

2、 In doing so , They should transfer the missing samples to the orthogonal line ;

3、 Each verifier can be on line / Transfer between columns 4 Missing samples （ about 55,000 An online verifier ensures complete reconfiguration ）

3、3 Danksharding DA sampling （ Malicious majority security ）

1、 Future upgrades

2、 Each full node checks the... On the block matrix 75 Random sample ;

3、 This ensures the probability that an unavailable block will pass < 2 ^(-30);

4、 bandwidth 75*512 B / 16s = 2.5 kb/s;

Four 、 Summarize the advantages and disadvantages

advantage ：

1、 Simple design ：

（1） No partition board infrastructure is required ;

（2） There is no need to track fragmentation blob confirm ;

（3） No additional builder infrastructure （ Synchronous payments from the execution layer replace this ）

（4） The existing executive fee market infrastructure can be used ;

2、 The tight coupling between the execution chain and the shard ：

（1） bring rollup The design is simpler ;

（2）zkRollup Synchronous calls between and the execution chain are possible ;

3、 There is no need for separate PBS;

4、 As the data is immediately verified by the verifier set 1/32（ Instead of the old partition scheme 1/2048） confirm , And in one epoch Period added to the complete set of verifiers , The ability to resist bribery has been enhanced ;

5、 Thanks to the 2d programme , All the nodes （ There is no need to run the validator ） Will be able to pass 75 Samples （2.5 kb/s） Ensure data availability , instead of 30*64=1920 Samples （60 kb/s）;