Proto Danksharding Resource Limits Explained Clearly

EIP-4844 introduces blobs. This new transaction type creates a dedicated, low-cost data channel separate from calldata, directly addressing the primary cost driver for Layer 2s like Arbitrum and Optimism.

The limit is 3 blobs per block. This is a soft, consensus-level target that provides a predictable, scalable data budget for rollups, unlike the variable and congested calldata market.

Blobs are ephemeral. Data persists for ~18 days, sufficient for fraud proofs and data availability sampling, which reduces long-term storage burdens compared to permanent calldata.

Evidence: Current calldata can consume over 90% of a block's gas. EIP-4844 is projected to reduce L2 transaction fees by 10-100x by decoupling data costs from EVM execution gas.

Calldata was a subsidy. Ethereum historically used execution-layer calldata for rollup data, creating a massive, mispriced state growth subsidy that threatened long-term security and fee predictability for chains like Arbitrum and Optimism.

Blobs are a separate resource. EIP-4844 introduces blob-carrying transactions with a dedicated fee market, isolating rollup data costs from EVM execution. This prevents congestion from an NFT mint on Ethereum from spiking fees for an Arbitrum user.

The target is 3 blobs per block. This initial limit, a precursor to full Danksharding's 64, provides ~0.375 MB of dedicated data space per slot, establishing a predictable base for rollup scaling without overloading consensus clients.

Evidence: Post-EIP-4844, L2 transaction fees on Arbitrum and Base are now dominated by execution costs, not data posting costs, which fell by over 90%. The blob fee market operates independently, proven during the recent EIP-4844 gas spike event.

Blob gas is separate. It exists on a distinct fee market from standard execution gas, preventing L2 data posting from congesting user transactions. This separation is the core mechanism for predictable L2 costs.

Blobs are ephemeral. Data blobs persist for ~18 days, not forever. This temporary storage enables the 1 MB per block target while keeping node storage requirements manageable for clients like Nethermind and Erigon.

Targets enforce scarcity. The system uses a target blob count and a maximum blob count per block. Fees adjust exponentially based on deviation from the target, creating a self-regulating market for block space.

Evidence: Post-EIP-4844, the average blob gas price has remained near 1 wei, while execution gas spiked during memecoin frenzies, proving the markets are decoupled. L2s like Arbitrum and Optimism now have stable data costs.

Proto-danksharding is a data availability layer. It introduces blob-carrying transactions and EIP-4844 to provide cheap, temporary data storage for rollups like Arbitrum and Optimism, decoupling data costs from execution gas.

Full danksharding requires data availability sampling. Clients will verify data availability by randomly sampling small chunks of the blob data, enabling secure scaling without downloading the entire dataset, a concept pioneered by Celestia.

The final bottleneck is execution. Even with infinite data bandwidth, a single execution thread remains limited. The endgame is a multi-threaded EVM or native cross-shard execution, moving beyond the current rollup-centric model.

Evidence: Current Ethereum processes ~50 TPS; full danksharding targets 100,000+ TPS for data, but execution scaling via zkEVM parallelism or Ethereum sharding is the unresolved challenge.