Layer 1 rookie Sei Network, what are the advantages of latecomers in terms of technical mechanism?

Author: Kylo@Foresight Ventures

The emergence of blockchain technology and its ecology provides many opportunities for innovators, developers and users. But so far, issues related to scalability, transaction speed, and front-running transactions have hindered Web3 from moving towards true mass adaptation. The emergence of Sei is to solve the above problems.

This article mainly expounds the superiority of the Sei mechanism through the description of the basic mechanism of Sei. This report is divided into several specific parts: "Introduction to Sei's Mechanism", "Sei's Flourishing Ecology", "Comparison of Sei with other layer 1" and "Sei's Unique Advantages in Trading".

Sei: Mechanism Introduction

Sei is a universal Layer1 network designed to solve various problems commonly found in layer1. It operates through the Twin-Turbo consensus mechanism and uses transaction parallelization to achieve fast transaction confirmation, high throughput and scalability. This innovative approach makes Sei a versatile and powerful platform that effectively bridges the gap that exists between decentralization and high performance.

Twin-Turbo Consensus

Sei is a high TPS parallel Layer1, which is similar to DyDx's Tendermint mechanism at the consensus level but different. In order to better understand the mechanism of Sei, we need to understand the traditional block generation mode. The generation of blocks is realized by validators by packaging transactions in their own memory pools and reaching consensus through the entire network. Each network validator has its own memory transaction pool, referred to as mempool. When a transaction occurs, the user will submit the information related to the transaction to a full node (the full node also belongs to the verifier), and the full node will send the relevant information to other nodes in the whole network. This process is gossip . After other nodes receive and verify the transaction information, they will add the transaction to their mempool. The proposer, that is, the person who builds the block will sort the transactions from his own mempool, produce the block, and broadcast the block information to the whole network, and the block information will be verified by other verifiers. After the verification is correct and a consensus is reached, other verifiers will accept the full block information from the proposer.

FIG.1.dumb block propagation

From the above mechanism, it can be found that there may be room for optimization in two steps:

• Since each verifier has a mempool, the transaction data in the new block may already exist in their own mempool. Other validators can generate blocks themselves without waiting for the proposer to send them detailed block data
• The confirmation of the block needs to carry out the process of block proposal, validator voting, reaching consensus and broadcasting the block in sequence. Since the above steps are performed serially, there is no way to speed it up. Block performance can be significantly improved if these several steps are processed in parallel

Sei has made optimizations at the consensus level based on the above two points. It defines its consensus mechanism as Twin-Turbo Consensus, which simply means a smarter block propagation method and an optimistic block production mechanism.

The first turbo needs to solve two problems:

• Let each validator's mempool contain all transaction information as much as possible
• Other verifiers can quickly know the transaction information packaged by the block proposer

It is an ideal state for all mempools to contain all transactions, but it is difficult to achieve in reality. So Sei made a remedy. The block proposer will break the block and send it to the whole network when proposing the block, and send a hash of all transactions in the block at the same time. Other verifiers look for corresponding transactions in their own mempool according to the hash. If some transactions are missing, they will find some missing transactions from the block fragments sent by the block proposer to realize block reconstruction. In this way, Sei Network reduces the time required for other verifiers to synchronize block information.

FIG.2.intelligent block propagation

Sei's second turbo uses an optimistic block generation mechanism. "Optimistic" once means that the majority of block proposers will not make mistakes by default. Under an optimistic assumption, the validator can synchronously process block data while performing prevote and precommit, that is, first write the block data proposed by the proposer into the cache, and if the block is verified, it can directly import the cache data. Instead of serially waiting for prevote and precommit to pass. Sei Network realizes the function of reducing transaction delay and improving blockchain performance through the above-mentioned Twin-Turbo Consensus.

Fig,3. comparison between “normal” and “intelligent”

Parallelization of transactions

Transaction parallelization is a common way for Layer 1 such as Solana and Aptos to improve throughput. But transaction parallelization has another meaning for Sei Network.

Sei Network, like DyDx's V4 version, puts the matching engine of the order book on the verifier node, and each verifier needs to maintain a mempool. When any verifier is selected as a block proposer, it needs to match transactions directly through the built-in matching engine and propose a block. For DyDx, the mempool only has transactions from DyDx; for Sei Network, due to the nature of its layer1 network, the transactions stored in the mempool come from various protocols on Sei. Most of these transactions from different protocols are not related to each other. If serialization is still implemented, Sei's various order book protocols will be in a state of competition for block space, which is not conducive to the development of the overall ecology.

Therefore, in summary, Sei's parallel design is actually to allow projects on Sei to run without interfering with each other, while also increasing the throughput of the system.

One of the main problems faced by transaction parallelization is the interconnection of transactions. Interrelated transactions can only be realized through the serialization of transactions. For example, the mint process of NFT needs to ensure that the minted NFT is not fully mint, so it can only be done serially. Therefore, how to distinguish between independent transactions and related transactions is a problem to be solved in the transaction parallel system. The UTXO model is one of the common ways to implement parallel transactions, and Sei uses DAG (Directed Acyclic Graph) technology to achieve it. In fact, DAG can be simply understood as a polyline with direction. The junction of the polyline is each transaction, and the two transactions connected to this transaction are related transactions of this transaction. Sei will set up a DAG for all transactions on the entire network to identify related transactions.

Fig. 4. Sei's transaction parallelization

MEV Prevention 、Order Bundling and Oracle Pricing

Sei's MEV Prevention function is mainly to prevent block proposers from maliciously extracting MEV when matching transactions and building blocks, and the way to achieve it is through Batch Auction. Batch Auction is Cowswap's solution to the MEV problem faced by AMM transactions. It packs the same type of transactions within a period of time into Batch and executes them uniformly. All transactions in the Batch have no order of execution and have the same execution price, thus avoiding the occurrence of frontrunning.

Order Bunding is a mechanism designed for market makers. Market makers can update the state of all order books with one trade instead of one by one. This means that market makers can quickly adjust their risk exposure in different order books at low cost.

In terms of oracles, Sei has a built-in oracle system that provides quotation services for asset prices within the ecosystem. Its main implementation method is to introduce the price quotation of the oracle machine into the consensus process. When each block is generated, all verifiers need to give their own quotations for asset prices and reach a consensus on the quotations. Therefore, the asset price of the entire Sei will be updated every block time.

The development status of Sei ecology

As of now, Sei's thriving ecosystem has included more than 150 projects across multiple web3 fields, including social, NFT, games, and DeFi.

Fig.5. Six's echo

Top projects in the Sei ecosystem include Fable, Dagora, and Fuzio, among others. Starting from Sei's special mechanism, they are exploring gameFi, NFT and DeFi. Developers choose Sei mainly for the following reasons:

• Scalability: Sei's high-performance architecture can handle thousands of transactions per second, which is suitable for dApps that require high throughput and low latency
• Low transaction fees: Sei's low transaction costs will incentivize developers and users to actively interact on Sei
• Ecological resources: Sei provides developers with rich development tools and resources, and opens community forums. At the same time, Sei also has the support of well-known investors and mature blockchain projects
• Interoperability and composability: Sei interacts seamlessly with other networks, enabling the composition of various protocols and applications.
• Security: Sei's consensus mechanism can ensure the security of transactions, allowing developers to focus on application development
• Community: Sei's growing number of users and active community are very attractive for developers
• Grants Ecological Fund: 120 million Sei Ecological Fund can help projects expand faster

Sei provides developers with a full-featured platform, allowing them to focus on exploring new possibilities for dapps through Sei's special mechanisms. In addition, Sei also guides the liquidity of funds on the chain through the "Liquidity Alliance Program", and provides web3 users with exposure to various projects through its huge community. At present, users on the Sei chain are relatively active. According to the test network data, more than 100 million transactions have occurred on Sei, with more than 5 million wallet addresses. Active user behavior on the chain will provide huge development potential for the Sei ecosystem.

DEX on Sei: Taking into account high performance and low cost

Speed and scalability are among the hallmarks of Sei, but it also has flexibility and adaptability for developers. Developers are free to build various applications on Sei. With its high throughput, low transaction fees, and fast finality, Sei provides an ideal infrastructure for building next-generation Web3 applications.

One of Sei's areas of strength lies in decentralized exchanges (DEX). In general, DEXs are vulnerable to blockchain performance issues. When the number of transactions per unit time surges, the problems faced by DEX due to blockchain congestion will become more prominent. High transaction fees, long transaction closing times, and poor scalability often hurt user experience and reduce profitability.

Sei proposes a proven method for the above market. It introduces an on-chain matching engine to make transactions on the chain more efficient by taking advantage of its characteristics of fast certainty, high throughput and low transaction costs. This means better performance and lower transaction costs for decentralized exchanges, making them potentially more competitive alternatives to centralized exchanges.

DEX on Sei has several advantages over Serum and DyDx. The problem with DyDx is that the mechanism design of the entire chain only serves one application, and lacks the living soil of other DeFi protocols, so it also loses the space for liquidity sharing and mutual combination between protocols. It must be connected with off-chain DeFi applications through some cross-chain means to realize the so-called DeFi composability; the problem Serum faces is that even if it has abundant liquidity and a huge ecology, it can easily realize the combination between protocols. Its stability will be greatly affected by the interference of non-trading activities. However, the Layer 1 design of Sei Network can solve the problems faced by DyDx and Serum. In short, Sei Network is characterized by the decentralized off-chain matching function and the composability of DeFi protocols, which gives Sei a strong advantage at the transaction level.

Comparative analysis of Sei and other layer1

The current Layer1 presents a state of blooming flowers, and each platform has its unique functions, advantages and limitations. To better understand Sei's strengths, we compared Sei with well-known platforms such as Sui, Aptos, Solana, and Ethereum.

May be

The key advantage of Sei is that its special network structure design solves the problems of scalability, transaction speed and front-running that are common in other blockchains. Sei leverages its ingenious consensus mechanism to achieve high throughput, fast finalization (as fast as 0.5 seconds), and scalable functionality. Sei's transaction parallelization design further enhances these capabilities, enabling Sei to handle a large number of transactions.

Sui

Although Sui has its unique advantages, compared with Sei, its scalability will face a bottleneck as its transaction speed increases. Although Sui also focuses on decentralization, its consensus mechanism is different from Sei in terms of flexibility, and it does not have the freedom of Sei in the selection of verifiers.

Solana

Similar to Sei, Solana will parallelize the production of blocks. It achieves high transaction speed and low transaction cost through POH, a unique timestamp system. However, its focus on performance necessitated compromises on decentralization. Due to the centralization problem, the stability of Solana will be affected by special circumstances.

Ethereum

Ethereum is currently the layer1 with the richest dapp ecosystem. Especially in the field of DeFi, it ranks first in terms of TVL and combinable attributes between protocols. However, Ethereum is currently facing high transaction fees and scalability issues, and its current solution is to distribute traffic through rollup. But the migration of applications and TVL between layer1 and layer2 still takes a long time. In conclusion, while Sui, Solana, and Ethereum all bring unique features and benefits, Sei still stands out due to its novel design and powerful performance. Sei's blockchain architecture is optimized for speed, scalability, and security without compromising decentralization. In the future, it may even surpass mature platforms such as Solana in terms of performance.

Fig.6. Comparative Analysis

Sei's Future

The potential of blockchain technology is enormous. By providing a highly scalable, secure and user-friendly environment, Sei paves the way for mass adaptation of blockchain technology. But from the perspective of the entire layer1 and layer2 competition pattern, Sei still faces fierce competition from other Layer 1 and Layer 2. Although Sei's mechanism and testnet data have allowed Sei to start, its subsequent prosperity still requires continuous cultivation of talents within the ecology and promotion of the community.

References