Proto-danksharding, specified in Ethereum Improvement Proposal 4844, is a change to the Ethereum blockchain that introduced a new transaction type carrying large packets of data called blobs. Blobs give rollups and other layer 2 networks a cheap place to publish the transaction data they must make available to the public, which was previously posted as expensive calldata. The change went live on Ethereum mainnet in March 2024 as part of the Dencun upgrade.
The name combines "proto", meaning a first step, with danksharding, the longer-term scaling design named after Ethereum researcher Dankrad Feist; the proposal was also nicknamed after researcher Protolambda, one of its authors. It belongs to the part of the Ethereum roadmap known as The Surge, which aims to scale the network through rollups rather than by raising the capacity of the base layer for ordinary transactions.
The problem it addressed
A rollup executes transactions off the main chain and posts compressed transaction data back to Ethereum so that anyone can reconstruct its state and, for an optimistic rollup, challenge an invalid result. Before EIP-4844 this data had to be included as calldata, which is stored permanently by every node and priced in the same gas market as all other activity. Data publication was the largest cost for most rollups, and it rose whenever the main chain was busy.
Rollups need the data to be available for long enough for anyone to check it, not forever. EIP-4844 separates that short-term availability need from permanent storage.
How blob transactions work
EIP-4844 defines a new transaction type that can carry one or more blobs. Each blob holds 4,096 field elements of 32 bytes each, so its size is:
That is 128 KiB per blob. The main features are:
- Data outside the execution layer. Blob contents are propagated and stored by the consensus layer and cannot be read by smart contracts. The execution layer sees only a short versioned hash that commits to each blob.
- KZG commitments. Each blob is committed to with a KZG polynomial commitment, a cryptographic technique that lets anyone verify that a given piece of data belongs to the blob. A new precompiled contract lets smart contracts, such as a rollup's verifier, check such proofs against the versioned hash. The scheme relies on parameters produced in a public multi-party ceremony held before the upgrade.
- Temporary storage. Consensus nodes are required to keep blobs for only about 18 days, after which they may delete them. The commitment remains on chain permanently.
The blob fee market
Blobs are priced in a separate unit called blob gas, with its own base fee that adjusts block by block in a way modelled on EIP-1559. When blocks contain more blobs than a target number, the blob base fee rises; when they contain fewer, it falls. The blob fee is burned, like the ordinary base fee. Because demand for blobs is separate from demand for normal block space, a surge in trading on the main chain does not directly raise the cost of rollup data.
At launch the target was three blobs per block and the maximum six. The Pectra upgrade in 2025 raised them to a target of six blobs and a maximum of nine under EIP-7691. After Dencun the fees paid by users of major rollups fell sharply, since blob space was initially underused and its base fee stayed near the minimum for long periods.
Path to full danksharding
Proto-danksharding introduced the transaction format, the fee market and the commitment scheme that full danksharding will use, but every consensus node still downloads every blob. Full danksharding aims to raise blob capacity much further by letting nodes check availability through data availability sampling, downloading only small random pieces of each blob. This is a form of sharding applied to data rather than to execution, and it is being introduced in stages by later upgrades.
Effects and criticism
- Cheaper rollups. The upgrade made posting data to Ethereum much cheaper and supported growth in activity on ZK-rollups and optimistic rollups.
- Value capture debate. Low blob fees reduced the amount rollups pay to Ethereum, prompting debate over whether the base layer captures enough value from the activity it secures.
- Data retention. Because blobs expire, anyone who needs historical rollup data must rely on archives kept by rollup operators, indexers or other third parties.
External links
Ref 1070001357