Roll-up Economics Update: Potential Impacts of EIP-4844

Brought to you by 0xfan

TLDR：

• We have conducted calculations using two different methods to assess the potential reduction in gas fees, TPS (transactions per second), and the capacity for accommodating Rollups after the implementation of EIP-4844.

• Based on our estimation, EIP-4844 has the potential to accommodate significantly more Calldata, ranging from 38 times to 192 times more. This corresponds to scenarios where the Calldata sizes are 10KB and 2KB, respectively. As a result of being able to accommodate more Calldata within the same block, the cost per unit of Calldata would also experience a proportional decrease.

• Assuming uniform Calldata size of 2KB for each Rollup, EIP-4844 would only be capable of accommodating a maximum of 384 Rollups.

• Under normal circumstances (i.e., when the block is at the target size), EIP-4844 would help Ethereum achieve a TPS of 175, with a maximum of 350.

• Contrary to common belief, relying solely on EIP-4844 is insufficient for Ethereum to achieve significant scalability improvements.

• Utilizing alternative DA Layers (e.g., Celestia) or DAC (e.g, zkPorter), improving the compression rate of L2 transaction data, and increasing the proportion of zk Rollups all have critical implications for further scalability of Ethereum.

Proto-danksharding, also known as EIP-4844, proposes to implement the majority of the logic and rules that Danksharding may use in the future. Currently, the high transition fee on L2 is mainly due to the high storage costs on L1. To address this issue, EIP-4844 introduces a new data type called Blob, which is cheaper and larger than calldata, providing an alternative way for rollup data storage.

With the upcoming EIP-4844, L2 sequencers can expect higher profit margins. This is because the sequencer is responsible for batching transactions into L1 and paying for the data fee, and the L1 data costs paid by the sequencer are set to decrease substantially. The low transaction fee could potentially generate more MEV by increasing the number of order flows on L2.

EIP-4844 is planned to be included in Ethereum's next upgrade, called the Cancun upgrade. However, there is currently no official release date. The Ethereum Foundation Research Team has stated that, optimistically, the upgrade could be released by the end of October. However, a more realistic expectation would be sometime around Q1 2024.

But to what extent can EIP-4844 reduce transaction fees? The transaction fees on L2 are composed of two primary components:

• Rollup Costs: The cost of batching transactions, submitting them to Ethereum and storing in Ethereum.

• Execution Costs: The cost to run the transaction on L2

L2 Transaction Fee = Rollup Costs + Execution Costs = [ L1 Gas Price * (Calldata + Fixed Overhead ) ] + [ L2 Gas Price * L2 Gas Used ]

Taking Optimism as an example, currently, almost 80% of the total transaction fee is due to the storage cost (i.e., Calldata Cost) on L1. We temporarily ignore the impact of other fees, and propose two methods to estimate the potential reduction in L2 transaction fees after EIP-4844.

Based on the current design outlined in  EIP-4844 , after the implementation of EIP-4844, the size of each Blob will be 128KB, and each Blob consumes 131,072 gas. Therefore, on average, each byte of Blob data would consume 128 * 1024 / 131072 = 1 gas. In comparison, currently storing a single byte of Calldata consumes 16 gas. This indicates that the storage cost of L2 transactions would decrease by a factor of 16.

However, this calculation only compares the storage cost per byte and does not take into account the total gas capacity of a block. As the total amount of gas a block can carry may change after EIP-4844, this approach somewhat underestimates the potential reduction.

The second method considers the block size and examines how many times the current Calldata can fit into different block sizes. Based on the current parameters, under the Target Block Size scenario, 3 Blobs (0.375MB) can fit into a block, and a block can accommodate a maximum of 6 Blobs (0.75MB). Considering that current Calldata occupies approximately 2-10KB per block , after EIP-4844, a maximum of 0.75 * 1024 / 2 = 384 times more Calldata can be accommodated.

However, as the block size increases from the Target Size to the Max Size, the gas price exhibits exponential growth. Therefore, for a more common case (i.e., when the block is at the target size), EIP-4844 can accommodate 38x - 192x more Calldata, corresponding to the cases of 10KB and 2KB Calldata, respectively. Due to the increased capacity of Calldata within a block, the storage cost of Calldata would correspondingly be decreased. Consequently, the cost of L2 transactions would be reduced by the same factor.

Furthermore, assuming uniform Calldata size of 2KB for each Rollup, EIP-4844 would only be capable of accommodating a maximum of 384 Rollups. This falls short of the thousands of Rollups that many people envision.

Following this, we can also derive the order of magnitude of TPS that Ethereum can achieve after EIP-4844. Currently, an average L2 transaction costs approximately 3,000 gas of Calldata on L1. Considering that each byte of Calldata carries a gas cost of 16, this indicates that each L2 transaction would be approximately 187 bytes on L1.

After EIP-4844, the target block size would be 0.375 MB, and Ethereum would generate a block every 12 seconds. Therefore, there would be 0.375 / 12 * 1024 = 32 KB of space available per second, which can accommodate 32 * 1024 / 187 = 175 transactions. Thus, under normal circumstances (i.e., when the block is at the target size), EIP-4844 would help Ethereum achieve a TPS of 175, with a maximum of 350.

While achieving a higher TPS can enhance efficiency, it is important to note that even after the implementation of EIP-4844, there is still a substantial gap when compared to Visa, which currently operates at a TPS of up to 1700. This discrepancy could still result in network congestion issues in both L1 and L2, especially in high-demand scenarios.

Therefore, relying solely on EIP-4844 is not enough for Ethereum to achieve greater scalability. Having a more cost-effective and efficient data availability solution to store more Calldata, such as a DA Layer like Celestia or a DAC like zkPorter, would still be crucial for achieving scalability even after the implementation of EIP-4844.

Lastly, the compression rate of L2 transaction directly affects the size of Calldata stored in L1. A higher compression rate results in lower required L1 fees. As zkRollup continues to advance, the amount of data that needs to be stored on L1 decreases, offering potential scalability for Ethereum. This is due to the fact that zkRollup, unlike optimistic Rollup, only necessitates the storage of state changes rather than the entire transaction.

Conclusion

In this article, We have conducted calculations using two different methods to assess the potential reduction in gas fees, TPS (transactions per second), and the capacity for accommodating Rollups after the implementation of EIP-4844. The results indicate that, assuming a uniform Calldata size of 2KB for each Rollup, EIP-4844 would only be able to support a maximum of less than 400 Rollups. This falls short of the envisioned thousands of Rollups that many people anticipate. Utilizing alternative DA Layers or DAC, improving the compression rate of L2 transaction data, and increasing the proportion of zk Rollups all have critical implications for further scalability of Ethereum.