Exploring the Rollup Ecosystem

The concept of rollups gained attention from the community after Vitalik Buterin announced Ethereum’s rollup-focused strategy in early 2020. After more than three years of development, the secure rollup system has demonstrated its ability to operate smoothly and reliably, securing billions of dollars in Total Value Locked (TVL).

Key Points

• Currently, the revenue from operating rollups is calculated by deducting operating expenses and L1 (Layer 1) costs from user transaction fees.
• While most rollups are functioning smoothly, certain aspects need improvement for enhanced fault tolerance and resilience.
• Rollup operators will become more specialized in making the network more flexible and adaptive, providing each specialized role with more opportunities for growth.
• Rollups, along with Data Availability (DA) and consensus, are exploring shared services that can improve various aspects of the process.
• DA is a highly competitive field with tremendous potential for creating value in the rollup technology stack. Ethereum is leading in this area and will continue to upgrade towards a new vision in the future.

Overview of Rollups

Rollups are off-chain scaling solutions aimed at increasing the throughput of the base layer (Layer 1) without altering the existing protocol. They allow computation, storage, and state transitions to be executed off-chain, external to Layer 1.

Rollups use two methods to verify accuracy and prevent fraud:

• Fraud Proofs: When a rollup submits a state root to Layer 1, a challenge period begins, during which anyone who wishes to contest its accuracy can submit evidence of fraud or error. If no challenges are made during this period, the state root is considered valid.
• Validity Proofs: Utilizes Zero-Knowledge Proof (ZKP) technology to generate valid proofs for state updates (state roots). These proofs are later verified on Layer 1 to ensure the state’s correctness.

State Root: A hash function captures the current state of the blockchain, including transaction balances, smart contract states, etc.

Transaction data also plays a crucial role in the rollup system. It can be used to compute and generate the correct state in cases of fraud or disputes. The classification of a rollup is determined by whether the transaction data is published:

• On-chain DA: Directly published on Ethereum
• Off-chain DA: Published on alternative solutions (outside of Ethereum) to save costs.

Generally, a “rollup” refers to a system that verifies states and publishes transaction data on Ethereum.

Figure 1: Classification of Rollups

Basic Rollup Economic Model

The economic model of rollups can be summarized as a model that includes the income, costs, and profits of three main groups: users, rollup operators, and the base layer.

Three Key Components of Rollup Systems

The rollup economic model consists of three core entities: users, rollup operators, and the base layer (Layer 1). Users: Users are participants in the rollup ecosystem who perform transactions and smart contract interactions. Rollup Operators: Rollup operators manage the infrastructure and overall operation of the rollup chain. They perform various tasks including:

• Transaction Sequencing: Organizing user transactions sent to the rollup in order, grouping them into batches, and periodically sending these batches to Layer 1.
• Transaction Execution: Storing and executing operations and updating the state of the rollup.
• Proposing: Proposers update the state root of the rollup on Layer 1 periodically.
• State Root Challenge: Submitting evidence of state root fraud and challenging the state root on Layer 1 (applicable only to Optimistic Rollups).
• Proving: Generating validations for each state root status update from the rollup to L1 (applicable only to ZK Rollups and Validium).

Note: These roles and tasks may vary between rollup projects, and some role names may not match this description. Base Layer (Layer 1): Rollups post transaction data and update the root state to Layer 1 after processing transactions and achieving consensus. Note: Optimistic chains/Validium do not post data to Layer 1.

Rollup Revenue, Costs, and Profit Model

The three roles mentioned above provide a basic framework for visualizing the revenue, costs, and profits of a rollup protocol.

Figure 2: Rollup Business Model

Rollup Revenue

Revenue from rollups primarily comes from two sources:

• MEV (Maximal Extractable Value): MEV is generated either within the rollup itself or through cross-chain MEV (both intra-domain and inter-domain).
• Transaction Fees: The total fees paid by users when interacting with the rollup.

Transaction fees are currently the only source of revenue for the rollup protocol, although MEV is available for use. Note: Some specific rollup projects like dYdX or Immutable X also offer gas-free transactions to incentivize platform use. These projects can charge fees on products or services, like trading.

Rollup Costs

Rollup costs are mainly divided into two categories:

• Rollup Operating Costs: These include various costs associated with running the rollup protocol.
• Layer 1 Costs (L1 Costs): Costs related to Layer 1.

Rollup Operating Costs:

Rollup operators incur various costs to operate the protocol. This includes expenses for hardware/VPS, utilities, internet, and maintenance of the rollup nodes.

Layer 1 Costs:

Rollups must pay Layer 1 costs:

• The cost of updating the state root on Layer 1.
• The cost of data publishing on Layer 1.
• ZK Rollups/Validium need to pay additional costs for Layer 1 ZK proof verification.

In most cases, Layer 1 costs constitute the largest proportion, especially when transaction data publishing accounts for most of the Layer 1 costs.

Most rollup protocols choose Ethereum as the base layer. Ethereum transaction data is published as call data, and costs are calculated using the following rates:

• 16 gas per non-zero byte.
• 4 gas per zero byte.

Call data is data transmitted alongside an Ethereum transaction, allowing users to send messages to other entities or interact with smart contracts.

Rollup Revenue

Rollups, at a minimum, charge users rollup fees based on Layer 1 and rollup operating costs, either directly or indirectly.

Simply put, the transaction fees of a rollup cover or exceed its expenses. Any surplus is the rollup’s profit.

Rollup Revenue = Rollup Fees — Rollup Operating Costs — Layer 1 Costs

Figure 3: Rollup Revenue

State of Rollup Technology Stack

The rollup technology stack can be described as a collection of various technologies used to build rollup applications. Since Vitalik Buterin announced Ethereum’s rollup-focused vision in early 2020, the rollup technology stack has grown significantly. Currently, rollup systems secure up to 10 billion dollars in TVL and offer excellent efficiency.

However, there are still many areas requiring further development, and two main concerns for rollup users are as follows:

• Centralization Risk: There can be centralization risks if the core development team performs crucial functions within the network.
• Technology Incompleteness Risk: Rollup solutions need to improve their technology stack to reduce user risks (Figure 4).

Figure 4: Top Rollup Projects’ Risk Analysis

From a positive perspective, competition will accelerate rollup development. With rollup projects exploring new ideas in transaction ordering policies, fee mechanisms, and shared services, the rollup technology stack is likely to become more sophisticated in the next 6–24 months.

Just as miners in Ethereum’s Proof of Work were distinguished as block creators and proposers, enhancing network flexibility and efficiency, rollup operators in Ethereum’s Proof of Stake will undertake more specialized roles. Thus, the growth space for each system role will expand.

When designing a rollup, developers can choose from various options with their own advantages and disadvantages in terms of cost, security, and integration with the Ethereum ecosystem.

So, how will rollup protocols evolve over the next 6–24 months?

Shared Services

Shared Services (SS) refer to common infrastructure or functionalities that can be utilized by various different rollup protocols for multiple benefits:

• Improve system resilience, ensuring normal operation even in case of errors or issues like offline nodes or network attacks.
• Facilitate cross-rollup interoperability.
• Enhance the security of rollup solutions through shared services.
• Offer various economic benefits like cost savings, economies of scale, and MEV extraction capabilities.
• Support technology through SDK rollup offerings.

In the most basic design, many rollup projects use Ethereum as the base layer to update states, verify truth or fraud proofs, and publish shared data. This is the most common rollup shared service.

The market will soon see other types of shared services targeting various aspects of rollup operation and offering benefits to participating rollup projects.

Rollup projects still face economic limitations due to fixed and variable costs related to settlement and transaction processing, including:

• Fixed Costs: Costs to send transaction batches to Ethereum even when there are no transactions in the rollup.
• Variable Costs: Costs that vary depending on rollup transaction activity.

Continuing to prioritize rollup operating costs includes:

• Transaction Sequencing → “Shared Sequencer”
• L2 Data Publishing → “DA” or “Shared Batch Publishing”
• Validity Proof Verification → “Proof Aggregation”
• Validity Proof Generation → “Proof Market”
• Cross-Rollup Communication → Cross-chain Messaging

From an objective perspective, shared services are just one development option for rollups. Before deciding on the development path, rollup developers must consider opportunity costs.

Apart from settlement and DA, major players in the rollup space like Arbitrum, Optimism, zkSync, StarkNet, and Polygon have decided not to participate in additional shared services to reduce technological dependence on external projects.

Instead, they can opt to develop their own shared services to build a stronger network environment and add value to the network. For example, Optimism developed Superchain, zkSync developed Hyperchain, and Polygon developed Polygon 2.0.

DA Layer: Ethereum vs. Alternative DA Solutions

The Data Availability (DA) layer refers to the ability of nodes to access and retrieve network data. Rollup protocols must publish transaction data to Layer 1 to reproduce off-chain states when necessary.

Ethereum rollup projects, especially Optimistic rollups like Arbitrum One and Optimism Mainnet, sometimes spend over 90% of their budget on DA. Major players are interested in the DA layer of the rollup technology stack, which is evaluated for its potential in value creation.

Rollup protocols typically utilize Ethereum as the DA layer. This helps to gain benefits from connections with the main crypto ecosystem, especially making ecosystem development easier for general-purpose rollups, but it also incurs significant DA costs.

Figure 5: Data Availability and Settlement Status of Top 10 Rollup Projects by TVL

Rollup projects can use a less secure and cheaper DA layer, but this may reduce developer and user interest.

Economically, using Ethereum as a DA layer balances cost efficiency and connectivity with the Ethereum ecosystem.

Figure 6: Advantages and Disadvantages of Ethereum DA and Off-Chain DA

Ethereum Proto-Danksharding (EIP-4844)

EIP-4844, also known as proto-danksharding, is a major Ethereum upgrade planned for implementation in Mainnet in Q1 2024.

This upgrade will provide dedicated storage for rollups to publish transaction data in ‘blob’ format. This data will be stored in the consensus layer for 18 days (4,096 epochs) before being removed from Ethereum.

EIP-4844 introduces a proposal for blob formats and a new “data gas fee market” to determine the price of blob transactions.

Blob data resources are not subject to the standard gas fee market formed by EIP-1559. Instead, prices are determined by the supply and demand of blob data.

Standard transaction prices remain unchanged, with call data costing 16 gas per non-zero byte and 4 gas per zero byte.

Only blob transactions can utilize both markets:

• EVM activities within the transaction are priced based on the standard gas fee market.
• Blob data within the transaction is priced based on the data gas fee market.

Rollups can publish transactions as call data using a unilateral fee mechanism (EIP-1559) or as blob transactions using a bilateral fee mechanism (EIP-1559 and EIP-4844).

The current EIP-4844 gas fee data mechanism, derived from EIP-1559, sells data storage space in blobs, integer units, each corresponding to 128kB.

• The standard unit is 3 blobs per block, i.e., 384kB (0.375MB). Using more than 3 blobs increases the cost of the next blob by 12.5%. Using less than 3 blobs decreases the cost by 12.5%.
• The maximum usage is 6 blobs/block, i.e., 768kB (0.75MB).

According to Arbitrum’s transaction data publishing data, the most data was published on March 23, 2023, the day the ARB token was launched.

Arbitrum consistently published approximately 3,398 batches, each about 100KB. Post-merge, Ethereum’s block time will be fixed at 12 seconds per block, generating about 7,200 blocks per day.

Based on this data, Arbitrum’s estimated data publishing demand on March 23, 2023, was about 47.2kB per block, an example of the storage limits EIP-4844 could impose on rollup protocols.

Currently, the data publishing demand of all rollup protocols fluctuates around 1.5 times that of Arbitrum’s data publishing demand of approximately 70kB per block as of March 23, 2023.

According to the current EIP-4844 fee mechanism settings, assuming maximum utilization of blob data space, blob prices will remain low, estimated at about 8 times the current demand, until the data publishing demand exceeds the standard of 3 blobs (about 384kB) per block. This level is expected to be reached within 1–3 years.

Based on the above reasoning, EIP-4844 should reduce batch publication fees for rollups by 65–90% depending on blob demand.

After EIP-4844, the Ethereum DA layer will impose ‘sufficiently low’ prices for rollups, so the balance between cost efficiency and connectivity with the Ethereum ecosystem may no longer be clear.

Alternative DA Solutions

While Ethereum remains the best choice for a DA layer, there are solutions and competitors (both direct and indirect) in the alternative DA space that could replace Ethereum.

Figure 7: DA Solutions for Rollups

New shared DA solutions optimize data availability using Layer 1 blockchains. Notable projects include:

• EigenDA: EigenDA is a DA layer built on the concept of re-staking and also leverages core ideas and libraries (Ethereum Danksharding codebase storage).
• Celestia: As a pioneer in the modular blockchain space, Celestia uses the Cosmos SDK to build an optimal blockchain specialized in data availability.
• Avail Project: A Layer 1 blockchain optimized for data availability, led by Anurag Arjun, co-founder of Polygon.

Customized DA solutions for specific rollup projects: zkSync’s zkPorter, Arbitrum’s Anytrust, Starkex DAC.

‘Traditional Layer 1’ blockchains upgrade themselves to make it easier for rollups to publish transaction data. For example, Tezos approaches this with ideas similar to Ethereum.

Considering Layer 1 data availability, most high-value financial applications may be sufficient with EIP-4844. Another market sector that alternative DA solutions could target includes non-financial applications requiring additional data and bandwidth.

Rollup Fee Model

Each rollup user transaction utilizes both Layer 1 and Layer 2 resources. Each network applies different fees for resource consumption. Thus, two essential components make up rollup transaction costs:

• Cost of utilizing resources on Layer 1.
• Cost of utilizing resources on Layer 2.

Rollup operators typically charge users fees for Layer 1 activities (in whole or part). Operators can customize their rollup resource measurement and pricing models to suit their needs.

Arbitrum, zkSync Era, and some other projects have introduced fixed minimum gas fees to ensure network security and efficiency while preventing network spam and maintaining competitive transaction fees.

• Arbitrum One’s fixed minimum gas fee is 0.1.
• Arbitrum Nova’s fixed minimum gas fee is 0.01.
• ZkSync Era uses a fixed minimum gas fee of 0.3.
• OP Mainnet implemented EIP-1559 with custom parameters post-Bedrock upgrade, building a Layer 2 fee market that fluctuates with the demand and supply of OP Mainnet block space.

transaction_gas_price = l2_base_gas_price + l2_priority_gas_price l2_transaction_fee = transaction_gas_price * l2_gas_used

l2_base_gas_price: The base gas fee within a block. If a block consumes more gas than the target, the base gas fee increases by 10%. If not, it decreases by 2% for less gas usage.

l2_priority_gas_price: A specially set priority gas fee for each transaction. Transactions are typically arranged and processed based on priority fees.

Figure 8: EIP-1559 Data of Op Mainnet

Transaction fees are essential for value creation in the rollup economic model. Rollup projects need a Layer 2 fee mechanism aligned with the project’s direction for efficient value extraction with reasonable trade-offs.

However, the Layer 2 transaction fee model is usually connected to the rollup’s transaction ordering mechanism, necessitating protocol modification or supplementation. For example, OP Mainnet developers added a mempool design to the protocol to fully implement EIP-4844.

Transaction ordering policy refers to how blockchain transactions are sorted and included in blocks, such as first-come-first-served or first auction.

New transaction ordering policies will be implemented in future rollup protocols, and the rollup fee model will change (compared to the current Arbitrum fee model) to extract value more effectively.

Figure 9: Ordering Policy & Fee Mechanism

Layer 3: Rollups Customized for Specific Purposes

Layer 3 (L3) involves building rollups on top of existing rollup protocols. While L2 focuses more on scalability, L3 focuses on specific rollup use cases. The purpose of L3 rollups can be distinguished from L2 as follows:

• L2 is designed to scale general applications (general rollups).
• L3 is customized for unique applications or use cases (specific rollups/appchains).

Many leading rollup projects plan to provide SDKs allowing developers to build their own L3 solutions on top of L2 rollup protocols.

Most major rollup initiatives aim to deploy L3 on L2 for additional revenue sources. Examples include Arbitrum’s Arbitrum Orbit, Optimism’s Op Stack, zkSync’s Hyperchain.

Figure 10: Popular SDK for Layer 3

Conclusion

The rollup ecosystem has undergone remarkable growth. While most rollups operate stably and smoothly, they still face several issues related to centralization risks and incomplete technology. Fortunately, rollup developers are exploring ways to make the current rollup technology stack more efficient and safe, and to prevent single points of failure.