A new transaction type that lets Ethereum temporarily store large data blobs cheaply, specifically for Layer 2 rollups.
EIP-4844 is crucial for Ethereum's scalability roadmap because it dramatically reduces the cost of posting data for Layer 2 rollups like Arbitrum and Optimism, which previously had to store all their transaction data as expensive calldata on the main chain. By introducing blob storage that is automatically pruned after a few weeks and doesn't need to be executed by the EVM, the upgrade can reduce L2 transaction fees by 10-100x while maintaining the security guarantees needed for rollups to function properly. This makes Ethereum-based applications more affordable for everyday users and represents a critical stepping stone toward full danksharding, where Ethereum will eventually support even more data availability through multiple blob slots, ultimately enabling the network to handle millions of transactions per second across all its Layer 2s.
Introduce a new transaction format for "blob-carrying transactions" which contain a large amount of data that cannot be accessed by EVM execution, but whose commitment can be accessed. The format is intended to be fully compatible with the format that will be used in full sharding.
Rollups are in the short and medium term, and possibly in the long term, the only trustless scaling solution for Ethereum. Transaction fees on L1 have been very high for months and there is greater urgency in doing anything required to help facilitate an ecosystem-wide move to rollups. Rollups are significantly reducing fees for many Ethereum users: Optimism and Arbitrum frequently provide fees that are ~3-8x lower than the Ethereum base layer itself, and ZK rollups, which have better data compression and can avoid including signatures, have fees ~40-100x lower than the base layer.
However, even these fees are too expensive for many users. The long-term solution to the long-term inadequacy of rollups by themselves has always been data sharding, which would add ~16 MB per block of dedicated data space to the chain that rollups could use. However, data sharding will still take a considerable amount of time to finish implementing and deploying.
This EIP provides a stop-gap solution until that point by implementing the transaction format that would be used in sharding, but not actually sharding those transactions. Instead, the data from this transaction format is simply part of the beacon chain and is fully downloaded by all consensus nodes (but can be deleted after only a relatively short delay). Compared to full data sharding, this EIP has a reduced cap on the number of these transactions that can be included, corresponding to a target of ~0.375 MB per block and a limit of ~0.75 MB.
| Constant | Value |
|---|---|
BLOB_TX_TYPE | Bytes1(0x03) |
BYTES_PER_FIELD_ELEMENT | 32 |
FIELD_ELEMENTS_PER_BLOB | 4096 |
BLS_MODULUS | 52435875175126190479447740508185965837690552500527637822603658699938581184513 |
VERSIONED_HASH_VERSION_KZG | Bytes1(0x01) |
POINT_EVALUATION_PRECOMPILE_ADDRESS | Bytes20(0x0A) |
POINT_EVALUATION_PRECOMPILE_GAS | 50000 |
MAX_BLOB_GAS_PER_BLOCK | 786432 |
TARGET_BLOB_GAS_PER_BLOCK | 393216 |
MIN_BASE_FEE_PER_BLOB_GAS | 1 |
BLOB_BASE_FEE_UPDATE_FRACTION | 3338477 |
GAS_PER_BLOB | 2**17 |
HASH_OPCODE_BYTE | Bytes1(0x49) |
HASH_OPCODE_GAS | 3 |
MIN_EPOCHS_FOR_BLOB_SIDECARS_REQUESTS | 4096 |
| Type | Base type | Additional checks |
|---|---|---|
Blob | ByteVector[BYTES_PER_FIELD_ELEMENT * FIELD_ELEMENTS_PER_BLOB] | |
VersionedHash | Bytes32 | |
KZGCommitment | Bytes48 | Perform IETF BLS signature "KeyValidate" check but do allow the identity point |
KZGProof | Bytes48 | Same as for KZGCommitment |
Throughout this proposal we use cryptographic methods and classes defined in the corresponding consensus 4844 specs.
Specifically, we use the following methods from polynomial-commitments.md:
Approximates factor * e ** (numerator / denominator) using Taylor expansion:
We introduce a new type of EIP-2718 transaction, "blob transaction", where the TransactionType is BLOB_TX_TYPE and the TransactionPayload is the RLP serialization of the following TransactionPayloadBody:
The fields chain_id, nonce, max_priority_fee_per_gas, max_fee_per_gas, gas_limit, value, data, and access_list follow the same semantics as EIP-1559.
The field to deviates slightly from the semantics with the exception that it MUST NOT be nil and therefore must always represent a 20-byte address. This means that blob transactions cannot have the form of a create transaction.
The field max_fee_per_blob_gas is a uint256 and the field blob_versioned_hashes represents a list of hash outputs from kzg_to_versioned_hash.
The EIP-2718 ReceiptPayload for this transaction is rlp([status, cumulative_transaction_gas_used, logs_bloom, logs]).
The signature values y_parity, r, and s are calculated by constructing a secp256k1 signature over the following digest:
keccak256(BLOB_TX_TYPE || rlp([chain_id, nonce, max_priority_fee_per_gas, max_fee_per_gas, gas_limit, to, value, data, access_list, max_fee_per_blob_gas, blob_versioned_hashes])).
The current header encoding is extended with two new 64-bit unsigned integer fields:
blob_gas_used is the total amount of blob gas consumed by the transactions within the block.excess_blob_gas is a running total of blob gas consumed in excess of the target, prior to the block. Blocks with above-target blob gas consumption increase this value, blocks with below-target blob gas consumption decrease it (bounded at 0).The resulting RLP encoding of the header is therefore:
The value of excess_blob_gas can be calculated using the parent header.
For the first post-fork block, both parent.blob_gas_used and parent.excess_blob_gas are evaluated as 0.
We introduce blob gas as a new type of gas. It is independent of normal gas and follows its own targeting rule, similar to EIP-1559.
We use the excess_blob_gas header field to store persistent data needed to compute the blob gas base fee. For now, only blobs are priced in blob gas.
The block validity conditions are modified to include blob gas checks (see the Execution layer validation section below).
The actual blob_fee as calculated via calc_blob_fee is deducted from the sender balance before transaction execution and burned, and is not refunded in case of transaction failure.
We add an instruction BLOBHASH (with opcode HASH_OPCODE_BYTE) which reads index from the top of the stack
as big-endian uint256, and replaces it on the stack with tx.blob_versioned_hashes[index]
if index < len(tx.blob_versioned_hashes), and otherwise with a zeroed bytes32 value.
The opcode has a gas cost of HASH_OPCODE_GAS.
Add a precompile at POINT_EVALUATION_PRECOMPILE_ADDRESS that verifies a KZG proof which claims that a blob
(represented by a commitment) evaluates to a given value at a given point.
The precompile costs POINT_EVALUATION_PRECOMPILE_GAS and executes the following logic:
The precompile MUST reject non-canonical field elements (i.e. provided field elements MUST be strictly less than BLS_MODULUS).
On the consensus layer the blobs are referenced, but not fully encoded, in the beacon block body. Instead of embedding the full contents in the body, the blobs are propagated separately, as "sidecars".
This "sidecar" design provides forward compatibility for further data increases by black-boxing is_data_available():
with full sharding is_data_available() can be replaced by data-availability-sampling (DAS) thus avoiding all blobs being downloaded by all beacon nodes on the network.
Note that the consensus layer is tasked with persisting the blobs for data availability, the execution layer is not.
The ethereum/consensus-specs repository defines the following consensus layer changes involved in this EIP:
On the execution layer, the block validity conditions are extended as follows:
Blob transactions have two network representations. During transaction gossip responses (PooledTransactions), the EIP-2718 TransactionPayload of the blob transaction is wrapped to become:
Each of these elements are defined as follows:
tx_payload_body - is the TransactionPayloadBody of standard EIP-2718 blob transactionblobs - list of Blob itemscommitments - list of KZGCommitment of the corresponding blobsproofs - list of KZGProof of the corresponding blobs and commitmentsThe node MUST validate tx_payload_body and verify the wrapped data against it. To do so, ensure that:
tx_payload_body.blob_versioned_hashes, blobs, commitments, and proofs.commitments hash to the versioned hashes, i.e. kzg_to_versioned_hash(commitments[i]) == tx_payload_body.blob_versioned_hashes[i]commitments match the corresponding blobs and proofs. (Note: this can be optimized using verify_blob_kzg_proof_batch, with a proof for a
random evaluation at a point derived from the commitment and blob data for each blob)For body retrieval responses (BlockBodies), the standard EIP-2718 blob transaction TransactionPayload is used.
Nodes MUST NOT automatically broadcast blob transactions to their peers.
Instead, those transactions are only announced using NewPooledTransactionHashes messages, and can then be manually requested via GetPooledTransactions.
This EIP introduces blob transactions in the same format in which they are expected to exist in the final sharding specification. This provides a temporary but significant scaling relief for rollups by allowing them to initially scale to 0.375 MB per slot, with a separate fee market allowing fees to be very low while usage of this system is limited.
The core goal of rollup scaling stopgaps is to provide temporary scaling relief, without imposing extra development burdens on rollups to take advantage of this relief. Today, rollups use calldata. In the future, rollups will have no choice but to use sharded data (also called "blobs") because sharded data will be much cheaper. Hence, rollups cannot avoid making a large upgrade to how they process data at least once along the way. But what we can do is ensure that rollups need to only upgrade once. This immediately implies that there are exactly two possibilities for a stopgap: (i) reducing the gas costs of existing calldata, and (ii) bringing forward the format that will be used for sharded data, but not yet actually sharding it. Previous EIPs were all a solution of category (i); this EIP is a solution of category (ii).
The main tradeoff in designing this EIP is that of implementing more now versus having to implement more later: do we implement 25% of the work on the way to full sharding, or 50%, or 75%?
The work that is already done in this EIP includes:
BeaconBlock verification and data availability sampling blobsBeaconBlock logic required for full shardingThe work that remains to be done to get to full sharding includes:
commitments in the consensus layer to allow 2D samplingThis EIP also sets the stage for longer-term protocol cleanups. For example, its (cleaner) gas base fee update rule could be applied to the primary basefee calculation.
Instead of putting rollup block data in transaction calldata, rollups would expect rollup block submitters to put the data into blobs. This guarantees availability (which is what rollups need) but would be much cheaper than calldata. Rollups need data to be available once, long enough to ensure honest actors can construct the rollup state, but not forever.
Optimistic rollups only need to actually provide the underlying data when fraud proofs are being submitted. The fraud proof can verify the transition in smaller steps, loading at most a few values of the blob at a time through calldata. For each value it would provide a KZG proof and use the point evaluation precompile to verify the value against the versioned hash that was submitted before, and then perform the fraud proof verification on that data as is done today.
ZK rollups would provide two commitments to their transaction or state delta data: the blob commitment (which the protocol ensures points to available data) and the ZK rollup's own commitment using whatever proof system the rollup uses internally. They would use a proof of equivalence protocol, using the point evaluation precompile, to prove that the two commitments refer to the same data.
We use versioned hashes (rather than commitments) as references to blobs in the execution layer to ensure forward compatibility with future changes. For example, if we need to switch to Merkle trees + STARKs for quantum-safety reasons, then we would add a new version, allowing the point evaluation precompile to work with the new format. Rollups would not have to make any EVM-level changes to how they work; sequencers would simply have to switch over to using a new transaction type at the appropriate time.
However, the point evaluation happens inside a finite field, and it is only well defined if the field modulus is known. Smart contracts could contain a table mapping the commitment version to a modulus, but this would not allow smart contract to take into account future upgrades to a modulus that is not known yet. By allowing access to the modulus inside the EVM, the smart contract can be built so that it can use future commitments and proofs, without ever needing an upgrade.
In the interest of not adding another precompile, we return the modulus and the polynomial degree directly from the point evaluation precompile. It can then be used by the caller. It is also "free" in that the caller can just ignore this part of the return value without incurring an extra cost -- systems that remain upgradable for the foreseeable future will likely use this route for now.
The base fee per blob gas update rule is intended to approximate the formula base_fee_per_blob_gas = MIN_BASE_FEE_PER_BLOB_GAS * e**(excess_blob_gas / BLOB_BASE_FEE_UPDATE_FRACTION),
where excess_blob_gas is the total "extra" amount of blob gas that the chain has consumed relative to the "targeted" number (TARGET_BLOB_GAS_PER_BLOCK per block).
Like EIP-1559, it's a self-correcting formula: as the excess goes higher, the base_fee_per_blob_gas increases exponentially, reducing usage and eventually forcing the excess back down.
The block-by-block behavior is roughly as follows.
If block N consumes X blob gas, then in block N+1 excess_blob_gas increases by X - TARGET_BLOB_GAS_PER_BLOCK,
and so the base_fee_per_blob_gas of block N+1 increases by a factor of e**((X - TARGET_BLOB_GAS_PER_BLOCK) / BLOB_BASE_FEE_UPDATE_FRACTION).
Hence, it has a similar effect to the existing EIP-1559, but is more "stable" in the sense that it responds in the same way to the same total usage regardless of how it's distributed.
The parameter BLOB_BASE_FEE_UPDATE_FRACTION controls the maximum rate of change of the base fee per blob gas. It is chosen to target a maximum change rate of e**(TARGET_BLOB_GAS_PER_BLOCK / BLOB_BASE_FEE_UPDATE_FRACTION) ≈ 1.125 per block.
The values for TARGET_BLOB_GAS_PER_BLOCK and MAX_BLOB_GAS_PER_BLOCK are chosen to correspond to a target of 3 blobs (0.375 MB) and maximum of 6 blobs (0.75 MB) per block. These small initial limits are intended to minimize the strain on the network created by this EIP and are expected to be increased in future upgrades as the network demonstrates reliability under larger blocks.
This EIP introduces a transaction type that has a distinct mempool version and execution-payload version, with only one-way convertibility between the two. The blobs are in the network representation and not in the consensus representation; instead, they are coupled with the beacon block. This means that there is now a part of a transaction that will not be accessible from the web3 API.
Blob transactions have a large data size at the mempool layer, which poses a mempool DoS risk, though not an unprecedented one as this also applies to transactions with large amounts of calldata.
By only broadcasting announcements for blob transactions, receiving nodes will have control over which and how many transactions to receive,
allowing them to throttle throughput to an acceptable level.
EIP-5793 will give further fine-grained control to nodes by extending the NewPooledTransactionHashes announcement messages to include the transaction type and size.
In addition, we recommend including a 1.1x base fee per blob gas bump requirement to the mempool transaction replacement rules.
Execution layer test cases for this EIP can be found in the eip4844_blobs of the ethereum/execution-spec-tests repository. Consensus layer test cases can be found here.
This EIP increases the bandwidth requirements per beacon block by a maximum of ~0.75 MB. This is 40% larger than the theoretical maximum size of a block today (30M gas / 16 gas per calldata byte = 1.875M bytes), and so it will not greatly increase worst-case bandwidth. Post-merge, block times are static rather than an unpredictable Poisson distribution, giving a guaranteed period of time for large blocks to propagate.
The sustained load of this EIP is much lower than alternatives that reduce calldata costs, even if the calldata is limited,
because there is no expectation that the blobs need to be stored for as long as an execution payload.
This makes it possible to implement a policy that these blobs must be kept for at least a certain period. The specific value chosen is MIN_EPOCHS_FOR_BLOB_SIDECARS_REQUESTS epochs, which is around 18 days,
a much shorter delay compared to proposed (but yet to be implemented) one-year rotation times for execution payload history.
Copyright and related rights waived via CC0.