Dencun: the Ethereum upgrade of March 2024 explained

Dencun was an Ethereum upgrade activated on March 13, 2024. It introduced a dedicated route for publishing rollup data alongside execution and consensus changes. Its main contribution was separating part of data availability costs from the ordinary execution market.

Distinguish transaction execution, publication of data needed for verification and long-term data retention. Dencun mainly changed the second. It did not make transactions free or guarantee a fixed fee across Layer 2 networks. Reducing that cost can improve rollup operating economics; the effect on users also depends on each network’s other costs.

Name and activation

Dencun combines Cancun, the execution-layer upgrade, and Deneb, the consensus-layer upgrade. Activation was scheduled for epoch 269568 on March 13, 2024 at 13:55 UTC. Node operators needed compatible clients for both layers.

ETH holders did not need to exchange their coins for a new asset. The upgrade changed protocol rules. It differs from The Merge, which completed the transition to proof of stake in September 2022, and Shapella, which enabled validator withdrawals in April 2023.

EIP-4844: data for rollups

A rollup executes operations outside the base layer and needs to publish data supporting verification or reconstruction under its design. Before Dencun, Ethereum transaction calldata was one common route. EIP-4844 added blob-carrying transactions with a separate fee market for those data.

Blobs are neither contracts nor new chains executing transactions in parallel. Their contents are distributed through the consensus layer and are not directly accessible as ordinary EVM data. Execution has cryptographic references and verification mechanisms defined by the proposal.

Protocol-required availability is temporary. Applications needing later historical retrieval must arrange retention. A cryptographic commitment can verify available data; it cannot reconstruct an unretained file.

Our EIP-4844 guide examines blobs, fees and availability in detail. Here the distinction helps explain their role within Dencun’s wider changes.

The nine included proposals

EIP-4844 introduced blob transactions and their gas accounting. EIP-7516 added BLOBBASEFEE, exposing the block’s blob base fee to the EVM. This differs from the execution gas base fee.

EIP-1153 introduced transient storage through TSTORE and TLOAD. Its state belongs to a transaction and is discarded at completion; it does not replace persistent storage. Contracts can use it to coordinate calls within an operation, subject to those rules.

EIP-5656 added MCOPY for memory copying. Using this EVM instruction depends on code and compilation tools. The upgrade did not automatically rewrite deployed contracts to use it.

EIP-4788 exposed beacon block roots through a contract accessible from execution. This supports verification involving consensus data with appropriate proofs and controls. It does not make arbitrary external information Ethereum-verified.

EIP-6780 changed SELFDESTRUCT. Outside the transaction creating the contract, it no longer deletes code and storage, while retaining transfer of the ether balance to the beneficiary. Creation and destruction in the same transaction preserve the earlier behaviour specified in the proposal.

EIP-7044 made voluntary exits signed under Deneb conditions remain valid across later upgrades. EIP-7045 extended the attestation inclusion window to the end of the epoch following the attestation’s epoch.

EIP-7514 capped validator activations at eight per epoch in that design. This is a historical Dencun parameter, not a complete description of entry rules after subsequent upgrades.

Why Layer 2 fees can fall

An L2 transaction price may include local execution, data publication, infrastructure and operator fee policy. Reducing one component can lower the total, but the proportion depends on the operation and network use.

In a hypothetical example, a transaction costs €0.10: €0.06 for data publication and €0.04 for other components. If publication falls to €0.01 with everything else unchanged, the total becomes €0.05. That is a 50% reduction, not the 83% reduction in the data component alone. This illustrates the calculation, not an observed fee.

Blob pricing also responds to demand and capacity. No universal, permanent savings percentage follows from Dencun. Measurement needs equivalent operations, dates, currencies and fee components.

What did not change automatically

Dencun did not remove L2 sequencers, upgrade permissions or bridge risks. Publishing data on Ethereum is one specific property; application guarantees also depend on validation, availability, proofs and exit mechanisms.

It did not promise low-cost base-layer operations either. Ordinary transfers and contract calls continue to compete for execution resources. Rollup data capacity and contract execution capacity are different dimensions.

What development teams should check

Verify that the compiler, testing environment and target network support the instructions used. Bytecode targeting a later hard fork may fail on a network that does not support it, even when both are described as EVM-compatible.

For L2 integrations, identify data publication and retrieval arrangements, user fee components and behaviour during sequencer or archive failure. Infrastructure improvements help when products understand those conditions and reflect them in operation.