In the rapidly evolving landscape of Rollups, practical updates are crucial for further scalability. Currently, Rollups process transactions off-chain, batch them, and submit them to Ethereum via calldata.
Table Content:
Calldata: A feature in Ethereum transactions allowing senders to include additional information.
Proportion of gas usage for sending data to Layer-1. Source: Dune @niftytable
Calldata costs represent the largest expense for Layer-2 projects (over 80% of total costs). This alone led Rollup projects to spend 15,000 ETH (approximately $34,000,000 USD) in Q2 2024.
Before prioritizing a rollup-centric roadmap, Ethereum pursued execution sharding, where each shard could execute different programs. However, due to limitations and risks, developers shifted towards data sharding and addressed Rollup scaling with “blobs.”
What are Blobs? Understanding Their Characteristics
Blobs are a new data type designed for temporary data storage for Ethereum applications, particularly Rollups. This format offers higher throughput at lower costs compared to traditional calldata.
Critically, consensus nodes process, store, and verify blobs independently of execution nodes on the execution layer. This means the EVM cannot access or store blob data, making Rollups the primary beneficiaries of this upgrade.
To accommodate short-term blob storage, Ethereum introduced a data gas market fee, pricing blob transactions separately from standard transactions.
Data storage is sold in whole blob units, each equivalent to 128 kB. Pricing is determined by market supply and demand:
- Standard rate: 3 blobs/block (384 kB). Above 3, the next blob’s price increases by 12.5%. Conversely, it decreases by 12.5% when usage is 3 or less.
- Maximum rate: 6 blobs/block (768 kB or 0.75 MB).
At these rates, blobs are projected to reduce batch publishing fees by 65-90%, depending on actual usage. Standard transaction fees remain unchanged, with calldata costing 16 gas/non-zero byte and 4 gas/zero byte. Only blob transactions utilize both gas markets. This dual-fee market allows rollups to choose between calldata with EIP-1559’s single-fee mechanism or blob transactions with both EIP-1559 and EIP-4844.
Blobs in Action: Real-World Impact on Ethereum
Following the successful Dencun upgrade, Layer-2 solutions like zkSync, Starknet, Arbitrum, Base, and Optimism swiftly implemented updates. Most reported significant transaction fee reductions (for sending ETH) due to lower data storage costs.
Average transaction fees post-Dencun on selected Layer-2 networks. Source: Dune @21co
Starknet and zkSync, being early adopters, initially dominated blob transactions on the mainnet. Subsequent upgrades by other projects have led to a more balanced distribution, with Base and Arbitrum now leading in blob usage (each at around 30-35%).
Blob usage by Layer-2. Source: Dune @hildobby
Layer-2s like Scroll, Linea, and Polygon zkEVM, yet to implement the upgrade, face higher transaction fees. Thanks to Dencun and blobs, Ethereum’s network load is reduced, reinforcing decentralization and security.
For application developers, projects sponsoring gasless transactions via Paymasters also benefit. In the past six months, these applications spent $1.8 million on gas subsidies, averaging $0.14 USD per user.
Proportion of sponsored Paymaster transactions (orange bars). Source: Bundle Bear
Post EIP-4844, projects would spend roughly $4,100 USD monthly to subsidize daily transactions (NFT minting, onboarding) for 10,000 users. Pre-EIP-4844, this would have cost $41,000 USD.
Conclusion
Blob transactions are a significant advancement for Ethereum. This update benefits not only Ethereum itself but also dramatically improves Rollup efficiency, reducing costs for users and applications.
Layer-2 projects can further optimize blob transaction usage by strategically timing transactions to coincide with full blobs and collaborating with other L2s to share blob space. Layer-2 solutions without blob transaction implementation will likely face disadvantages compared to those that have successfully integrated this feature.
