Rollup 2.0: The Battle for Decentralized Sequencers

I. Foreword

Time flies. According to Optimism, more than a year has passed, and according to Arbitrum, almost half a year has passed; issuing coins was only the first step in their long journey. During this period, Optimism upgraded Bedrock and launched the universal modular L2 stack OP Stack, which gave birth to star Rollups such as Base; Arbitrum is committed to exploring the application of L3 to promote Arbitrum Orbit.

Under the leadership of the two giants, the Rollup Track’s TVL once surpassed 10 billion US dollars and is currently stable at around 10 billion US dollars. Behind Rollups being relished as Ethereum’s “signature” scaling solution, they still have non-censorship-resistant and centralized attributes. Mainstream Rollup chains generally use official centralized sequencers. Although Rollup projects such as Arbitrum, Optimism, and StarkNet include the decentralization of sequencers in the roadmap, they have not been implemented in the short to medium term of planning. As the most important puzzle piece of Rollup’s decentralization puzzle, the decentralized sequencer has a very important strategic position for Rollup itself, and it is also the aspiration of the people.

According to L2Beat data, the TVL for the L2 track as of October 1, 2023

II. Rollup transaction fees

Before we understand what a sequencer is, let’s talk about the components of Rollup transaction fees. Rollup’s transaction fee is the gas fee that users incur in L2 transactions such as Arbitrum.

It mainly consists of 2 parts:
1) L2 execution costs
2) L1 data fee

L2 execution fee: the cost of transaction execution at L2 (each transaction initiated on the L2 chain is subject to an execution fee)
Transaction gas price = L2 base fee + L2 priority fee
L2 execution fee = transaction gas price * L2 gas usage

L1 data fee: The cost of posting L2 transactions to L1. Generally, L1 data costs more than L2 execution costs.

L2 transaction fee = L2 execution fee + L1 data fee

Net sequencer revenue = L2 transaction fee revenue - sequencer operating costs - L1 data fee

The centralized sequencer operated by the project party has a certain amount of pricing power (for example, L2 execution fees are a little higher, L1 data fees are a little higher), which is why several well-known Rollup projects are making a lot of money.

What is a sequencer?

A sequencer, as the name suggests, is a role responsible for ranking transactions. On the Bitcoin network, miners are responsible for ranking transactions; Ethereum is responsible for a collection of nodes, none of which play a fixed role, but rather a consensus mechanism to determine who is authorized to participate in sequential execution.

Currently, mainstream Rollups all run a single centralized sequencer. The user’s L2 transactions enter the mem pool (at this point, the transactions in the mem pool are in an unordered state), and the sequencer sorts and compresses the transactions into an ordered set of batches, and then sent to the DA layer of Ethereum.

Sequencer operation process

Does Rollup need a sequencer?

The answer is no. Transactions on Rollup can completely bypass the sequencer and be submitted to the L1 base layer. L1 is responsible for sorting and settlement, but they will also face higher gas consumption and longer transaction confirmation times.

The Rollup sequencer is similar to using a “fast track”, compressing hundreds or thousands of L2 transactions into a single L1 transaction, thereby greatly reducing gas costs. This is why today’s mainstream Rollups all run centralized sequencers, which provide users with lower gas and faster transaction confirmation, thereby improving the user’s transaction experience.

3. Centralized sequencer vs. decentralized sequencer

Centralize!

The advantages of centralization are very obvious. You can rank transactions however you want; there is no need to change the ranking, and there is no need to agree on the results of the ranking. This means that it has a very fast transaction confirmation speed, and the user experience is better;

However, centralization also gives the sequencer great autonomy to rank transactions. It can arbitrarily rank transactions to maximize its own arbitrage opportunities, seize MEV value, delay user transactions, and even completely censor users.

The sequencer can obtain MEV value by changing the transaction order within a single block; what is more harmful is that since the sequencer controls the sequencing of multiple blocks in a row, it is easy to execute MEV across blocks, causing large-scale attacks.

The above situations are all active misconduct by the sequencer. Some errors are not intended by the sequencer, but they still damage the user’s experience and rights. For example, the sequencer accidentally re-includes a token transaction that has already been spent in a soft promise and then sent to L1 for verification. There may also be cases where the transaction cannot be confirmed; it is also like a single sequencer going offline, causing the second layer to not be able to generate blocks properly, and the network is down for a long time.

Fish and bear are impossible, but optimizing Rollup’s performance must not come at the expense of decentralization and censorship resistance.

Decentralized?

If centralization is 1, then decentralization is much. There are differences in the implementation paths of different decentralized sequencer solutions, but their core concept is the same, which is decentralization.

The sequencer no longer has the right to centrally rank transactions. The role responsible for ranking is selected from a set of sequencer sets based on a specific election mechanism and rotates for a fixed cycle.

Decentralization prevents sequencers from continuously seizing MEV, and also prevents a single sequencer from reviewing users’ transactions. Coupled with the corresponding criminal punishment mechanism, it can also effectively regulate the behavior of the sequencer.

IV. Decentralized Sequencer Track Overview

After going around for so long, we finally got down to business. Decentralized sequencers. One is made by the Rollup project itself, and the other is implemented by a third party. Using a third party to implement a decentralized sequencer can actually also be called sequencing-as-a-service, and sequencing is a service.

Projects such as Espresso, Astria, SUAVE, and Radius all focus on decentralized sequencer solutions, and their implementation paths vary.

1. Espresso

Espresso Systems was initially a service provider focusing on privacy solutions. In March 2022, it was announced that it had received nearly $30 million in Series A financing with Electric Capital, Sequoia, and Blockchain Capital. Espresso Systems has now basically transformed into Espresso Sequencer, which specializes in providing decentralized sequencer services for Rollup.

Espresso’s funding

Under the Espresso sequencer’s ranking mechanism, L2 transactions generally experience the following life cycle:

1) Transactions made by users on the second layer are sent to the Rollup server (API);

2) Transactions enter the mem pool, and the sequencer (selected by HotShot consensus) sorts and includes transactions in a block;

3) The sequencer broadcasts the transaction. After reaching a HotShot consensus through other nodes, the block is issued, and the transaction is executed; the soft promise provides quick transaction confirmation

4) The sequencer sends and stores a consensus certificate (QC: Quorum Certificate) with the block promise containing the transaction in the L1 sequencer contract (proving that the block has reached soft finality through consensus);

5) The Rollup node that has executed the block sends the new Rollup state to L1 (at this point, zKrU requires proof of validity, and ORU opens the challenge period)

6) The L1 Rollup contract checks the validity of status updates by verifying the QC sent by the sequencer contract.

The life cycle of an L2 transaction under the Espresso sorting mechanism

This process seems obscure and difficult to understand; the simple way to understand it is:

The HotShot consensus selects one of a set of sequencers. It is responsible for sorting Rollup transactions and including transactions in a block; this block must be signed and agreed by other Rollup nodes (2/3 or more HotShot nodes agree) to be “final”, and then the relevant block promises and the new Rollup state root are submitted to the L1 base layer for verification.

“Finality” above is in quotation marks; this “finality with quotes” and finality without quotes is not a concept. “Finality” with quotes is to allow Rollup transactions to be confirmed more quickly, with less delay, and a better user experience; however, Rollup transactions ultimately require the L1 base layer to verify (zKru needs to verify proof of validity, ORU has to wait for the end of the challenge period). If there is no problem verifying that the transaction submitted by Rollup is OK, then Rollup’s transaction is truly final.

This means: if the L1 base layer verifies that the transaction is invalid, the relevant L2 block that has already been released will face a rollback. Therefore, “finality” is for transactions to be confirmed quickly, and finality is to inherit the security of Ethereum.

Espresso-free transaction sequencing architecture

Integrate Espresso’s transaction sequencing architecture

Espresso + eigenLayer

Espresso solved the sequencer rotation problem and the determination of the “finality” of the transaction based on the HotShot consensus, and solved the sequencer’s admission problem by introducing EigenLayer.

EigenLayer’s re-staking mechanism makes it possible for Ethereum stakers to simultaneously become Espresso sequencers, providing security for the HotShot consensus. Simply put, Ethereum node stakers can become Espresso Sequencers (ESQ) through EigenLayer’s re-staking mechanism. While receiving PoS node benefits, Ethereum stakers also capture the value of second-tier MEV.

Potential benefits for ETH holders = native network node rewards + L2 EVM + node rewards from other PoS chains (using the EigenLayer re-staking mechanism). The triple buff greatly empowers ETH.

EigenLayer’s Decentralized Sequencer Solution

Espresso is a general-purpose decentralized sequencer solution. In addition to EigenLayer, ecological cooperation projects include popular modular projects such as Arbitrum, OP Stack, Caldera, AltLayer, etc.

Espresso Ecological Cooperation Project Layout

2. Astria

Astria is positioned as a generic, permissionless decentralized sequencer, providing an out-of-the-box shared sequencer service for different Rollups. In terms of financing, Astria announced the completion of a $5.5 million seed round led by Maven 11 in April 2023. Co-investors include 1k (x), Delphi Digital, Lemniscap, Robot Ventures, etc. Although the financing scale is small, the institutional lineup is gorgeous.

Astria funding situation

Operating mechanism

The operation mechanism of the Astria decentralized sequencer is similar to the Espresso Sequencer. The purpose is to weaken the privileges of sequencers by delegating transaction sequencing rights. Let’s take a closer look:

For sequencer rotation, Astria proposed two rotation mechanisms: simple leader rotation (leader rotation) and Byzantine fault tolerance (BFT) consensus algorithm.

1) Leader rotation
A set is formed through an elected sequencer, and the sequencer set takes turns to sort Rollup transactions. This method prevents a single sequencer from continuing to monopolize the ranking rights of transactions for a long period of time, and solves the problem of continuous review of users to a certain extent.

Astria’s leadership rotation mechanism

2) BFT consensus algorithm
Similar to the leader rotation mechanism, the sequencer who takes the turn is responsible for ranking transactions, but 2/3 or more members in the sequencer set must agree on this ranking.

Each of the two methods has advantages and disadvantages: the former allows faster transaction confirmation, rapid block generation, and is close to that of a centralized sequencer. However, the trade-off point is that it is still difficult to restrain the sequencers in their turn from committing evil; using the BFT consensus is even less likely, and 2/3 of the sequencers in the set need to vote to reach a consensus before the block comes out. However, it took a certain amount of time to conduct a consensus vote, causing network delays.

Astria’s BFT consensus algorithm

3. SUAVE

SUAVE is a decentralized, plug-and-play shared sequencer solution built by Flashbots. As a general solution, SUAVE can provide memory pools and decentralized block construction for any L1/L2. The difference between SUAVE and the shared sequencer design described above is that SUAVE Chain itself is an EVM compatible chain, and transactions are ordered through block “bidding”.

SUAVE architecture

SUAVE’s architecture consists of 3 core components: a common preferences environment, an optimal execution market, and decentralized block construction.

1) Preferred environment
Preferences range from simple transactions to complex events. User preferences are reflected in mempools in the form of transactions, and the preferences environment is a public mempool that brings preferences together. The general preferences environment provided by SUAVE makes multi-chain user preferences open and transparent, eliminates poor information, and solves the problem of cross-chain MEV to a certain extent.

2) Execute the market
The execution market is a network where performers are responsible for monitoring SUAVE memepool and compete with each other, and competition drives them to provide the best execution for user preferences. It can be understood that all executors achieve user preferences through “bidding” and return as much MEV generated by user transactions to the user as possible.

3) Decentralized block construction
Finally, based on the collected preferences and the best execution path, the decentralized block construction network includes them in the block. At this point, the whole process of transaction discovery, transaction sequencing, and block generation has been realized.

SUAVE’s core components

4.Radius

Radius’s targeting is a shared ordering layer that doesn’t require trust. Unlike the implementation mechanisms described above, Radius ensures that Rollup transactions are ranked without trust by enabling encrypted mempools, thereby eliminating effective MEV and user transaction censorship.

In terms of financing, Radius announced the completion of the $1.7 million pre-seed round led by Hashed in June 2023, with co-investors including Superscrypt, LambdaClass, and Crypto.com.

Radius Funding Information

Decentralized sequencers based on consensus mechanisms such as Espresso and Astria reduce MEV and review risks to a certain extent, but at the cost of network scalability and time efficiency, they cause certain transaction confirmation delays (requiring consensus on transaction ranking). Furthermore, although transaction ranking is in a decentralized environment, since mempool related transactions are transparent, there is still room for mischief to seize MEV. Radius encrypts mempools, and relevant transaction information is not visible to the sequencer. The aim is to stop the sequencer from maliciously extracting MEV and reviewing transactions at the source.

Technical architecture

The Radius technology architecture can be divided into the following four functional layers: Radius (Radius), Execution Layer (Rollup), Settlement Layer, and Data Availability Layer.

1) Sorting layer

1. Users submit cryptographic transactions and certificates to the sequencer; 2. The sequencer verifies the certification and verifies the transaction; 3. The sequencer sorts the transactions, and the transactions cannot be decrypted until then; 4. The sequencer builds a block; 5. The sequencer submits blocks to Rollup for execution.

2) Execution layer

1. Rollup receives blocks from the sequencer and executes transactions in the order provided; 2. Rollup submits the transaction status and proof of status to the settlement layer.

3) Settlement layer

1. The settlement layer receives status and status certificates from Rollup, and is responsible for verifying the certification and determining the finality of the transaction; 2. The ordering layer verifies that the execution matches the order.

4) Data availability layer

The data availability layer stores data and ensures that it is available.

Radius’s major functional hierarchical architecture

Mempool’s encryption mechanism - PVDE

Radius uses “Practical Verifiable Delayed Encryption” (PVDE), an encryption scheme based on zero-knowledge proof, to create an encrypted mempool.

The specific process is as follows:

When a user submits a transaction to the sequencer:

1. The user generates a time-locked puzzle and a symmetric key;
2. The user uses a symmetric key to encrypt the transaction, and the encrypted transaction enters the mempool;
3. The sequencer sorts the encrypted transactions. The sequencer needs to unlock the time lock puzzle to obtain the decryption key;
4. The sequencer calculates the order promise before unlocking the time lock puzzle, and the settlement layer for the promise submission (used to verify that the sequencer submits transactions to Rollup in order).

Radius transaction encryption/decryption process

Decentralized solution for sequencers

Encrypted mempools ensure that the sequencer is untrustworthy, but the risk of a single point of failure still exists. If you run a single sequencer + encrypted mempool, a sequencer failure will cause the network to go down. In order to solve this problem, Radius proposed various decentralized sequencer implementation solutions, including secret leader election mechanisms, sequencer group segmentation mechanisms, etc.

Of course, Radius can also choose to refer to Espresso and Astria’s sequencer rotation mechanism, while making transaction sequencing decentralized and trustless.

Block space optimization

Through block space optimization, Radius aims to achieve the goal of protecting users while maximizing Rollup profits. Rollup uses a first-come-first-served service (FCFS) ranking mechanism. The advantage is that it can effectively prevent MEV, and the disadvantage is that the potential profit of block space auctions must be sacrificed.

To solve the transaction ranking dilemma described above, Radius divides the block space into 2 parts: top block space and bottom block space:

Among them, the top block space is dedicated to user transactions, encrypting user transactions to eliminate transaction ranking manipulation, thereby protecting users from harmful MEV and censorship risks; the bottom block introduces an auction-based transaction market where arbitrators can submit bundled transactions and their bids to the sequencer, and the sequencer will select the bundled transaction with the highest bid to be included in the block. This method can maximize Rollup’s profits.

The above is the current mainstream general decentralized sequencer solution. For Rollup, are you facing running a centralized sequencer or a decentralized sequencer? Integrate a third-party general-purpose sequencer solution or decentralize it yourself? What kind of technology is used to implement solutions to decentralize transaction sequencing? Weigh the pros and cons of multiple dimensions, etc.

5. The game between Rollups

Trade-off 1: Decentralized or not?

Several mainstream Rollups, such as Optimism, Arbitrum, zkSync, and Base, earn a lot of money by running centralized sequencers. Decentralization will inevitably involve profit sharing. Without considering the increasingly fiercely competitive landscape of the Rollup circuit, no one wants to share this sweet treat. But let’s say Rollup takes the lead in launching a decentralized sequencer. This is probably a huge traffic entry point, forming a demonstration effect in Rollup’s segmentation circuit, thus forcing other Rollup projects to decentralize their sequencers.

Trade-off 2: Integrating generic third party vs. in-house dedicated solutions

Generally speaking, there are 2 ways for sequencers to achieve decentralization: one is to use what others have done; the other is to do it yourself. Since third parties such as Espresso and Astria can provide Rollup with out-of-the-box decentralized sequencer services, Rollup itself can continue to focus on product differentiation and optimizing performance to improve its core competitiveness; in addition, integrated general-purpose decentralized sequencer solutions are also more conducive to achieving interoperability, bringing more possibilities, including cross-Rollup arbitrage. The downside of this solution may be that Rollup’s own native token cannot be effectively empowered.

If Rollup uses an in-house dedicated solution to implement a decentralized sequencer, this is the most time-consuming and expensive solution, but it is indeed the most effective way to enable Rollup’s native token. For example, the StarkNet project party can require the user to stake the native token of the agreement as a sequencer to participate in the Rollup transaction ranking, and charge a certain service fee to achieve value accumulation.

Trade-off 3: Which technology implementation solution is used to achieve decentralized ranking?

As mentioned earlier, there are many technical implementation solutions to achieve decentralized transaction ranking, including but not limited to solutions based on different consensus mechanisms, FCFS, block bidding, and encrypted mempools. Each technology implementation solution has advantages and disadvantages: based on consensus mechanisms, time efficiency will be limited, encrypted mempools cannot maximize Rollup profits, etc. Of course, you can also refer to Astria’s integration of 2 different technology implementations. The trade-off between the various technical implementations is an issue that all Rollup projects require careful consideration.

Write at the end

Although Optimism and Arbitrum, the leaders of the Rollup circuit, have now issued coins, this is probably just the starting point; competition in the true sense of the word may have only just begun. At least judging from current trends, decentralized sequencers must be a battleground for the military.

ZK Rollup projects are also quietly rising. In an increasingly competitive environment, taking a wrong step may cause irreparable losses. However, in the face of innovations affecting the life and death of the project, Rollups had no choice but to adapt to the general trend.

Disclaimer：

1. This article is reprinted from [mirror]. All copyrights belong to the original author [helloyyy]. If there are objections to this reprint, please contact the Gate Learn team（[email protected]）, and they will handle it promptly.
2. Liability Disclaimer: The views and opinions expressed in this article are solely those of the author and do not constitute any investment advice.
3. Translations of the article into other languages are done by the Gate Learn team. Unless mentioned, copying, distributing, or plagiarizing the translated articles is prohibited.