Data Availability

Data availability is the guarantee that the data behind a blockchain's blocks has actually been published so anyone can download it, verify the chain, and reconstruct its state. It sounds mundane but is one of the central problems in blockchain scaling: a block producer could publish a block header while withholding the underlying transactions, leaving the network unable to check validity or, on a rollup, unable to prove ownership of funds.

The issue is sharpest for layer 2s. A rollup's security promise — that users can always exit to the base chain even if operators turn hostile — holds only if the rollup's transaction data is available somewhere trustworthy. Rollups that post data to Ethereum inherit its availability guarantees; Ethereum's Dencun upgrade (March 2024) introduced "blobs" via EIP-4844, a dedicated cheap data space for exactly this purpose, which cut rollup costs dramatically. Systems that instead keep data with an off-chain committee ("validiums") are cheaper but reintroduce trust: if the committee withholds data, users may be unable to prove their balances. Specialized data availability layers such as Celestia, and Ethereum's own roadmap toward danksharding with data availability sampling — where light nodes verify availability by randomly sampling small pieces rather than downloading everything — aim to scale this guarantee.

A common misconception is that data availability means long-term storage; the guarantee is that data was published and obtainable when the block was made — Ethereum blobs, for instance, are pruned after roughly 18 days, by which time anyone who needed the data could have archived it.