Data Availability (DA)
Data Availability (DA) is the assurance that all the data a block producer publishes, which is needed to verify the state of a blockchain or Layer 2 rollup, is actually accessible to every network participant.
To check a block, a node needs the transactions inside it, not just a summary. If a block producer published a block header but withheld the underlying data, nobody could confirm the block was valid or spot fraud. Full nodes deal with this by downloading everything. Lighter clients can use Data Availability Sampling, where they randomly fetch small pieces of a block. Combined with erasure coding, this lets them become confident that the whole block was published without downloading all of it.
DA matters most for rollups. A rollup batches transactions off the main chain and posts compressed data back to Layer 1, which acts as its data availability layer. Because that data is public, anyone can reconstruct the rollup's state, challenge invalid batches, or withdraw funds even if the rollup operator stops cooperating. Validiums instead keep data off-chain, gaining scalability but relying more on whoever holds that data. Volitions let users choose between the two.
On Ethereum, adding dedicated data space called blobs was a step toward danksharding. The aim is to give rollups more room to post data without sacrificing decentralization.
A common misconception is that data availability means permanent storage. It only means the data was published and could be retrieved when needed for verification. Blob data, for example, is pruned after a limited period. Available data also does not prove transactions are correct. That is the job of fraud proofs or validity proofs.