# Let's talk about data-availibility

## Use case

### For Layer-2

- Smart rollups

- Validity rollups (aka zk rollups)

### A decentralised DB

Bandwith: ~ 10Mo/s

## Scalability of blockchains

- Computation
- Maintaining the state
- Bandwith

Rollups target the first two items, the DAL targets the third

## Data-availibility problem

Once the data are not on-chain, how can we guarantee they are still available?

:::warning
Difference between data publication and data retriviability
:::

## A sharding solution for Tezos (codename DAL)

- Another P2P network
- Slot producer produces data for the DAL
- Slot producer produces a commitment of the data and post it onto the L1
- Slot producer sends to the DAL shards of the original data
- Endorsers (aka attestors) connects to the DAL and fetch the shards they were assigned to by the L1
- Endorsers commits on the availibility of the data they were assigned
- The L1 aggregates the endorsers commitment and decides whether the data is available or not
- slot consumer may reconstruct the data

How to ensure data are still available after the attestation of endorsers? We use sampling (or something else)!

## Sampling

For the future

# For the users

## Attestor

## Developer

### Smart rollups

### Validity rollup

### Others

- Difference with IPFS