01

Does one blockchain need to do everything?

Early blockchains executed transactions, settled results, reached consensus on ordering, and stored transaction data all on one network, an approach often called monolithic. Increasingly, these roles are being split across separate layers in what is called a modular design.

Splitting roles lets each layer specialize and scale, but users need to know which layer is responsible for what, and what stops working if one of them fails. This guide explains the four roles, how they are divided, the security dependencies that result, and how withdrawal paths differ.

02

The four jobs of a blockchain

Execution processes transactions and updates balances and contract state. Settlement verifies those results, resolves disputes, and fixes the final outcome. Consensus is how participants agree on ordering and blocks. Data availability is the guarantee that transaction data is published so anyone can download it and check the results independently.

Ethereum.org's page on data availability explains that if block producers withheld data, other participants could not verify transactions, which is why the availability of data is central to security. In a modular design, these four roles can be assigned to different networks.

03

Who does what

The table compares how the roles are divided in three common designs.

Role division by blockchain design
RoleMonolithic chainRollup on EthereumRollup using an external DA layer
ExecutionSame chainRollupRollup
SettlementSame chainEthereumEthereum or another settlement layer
ConsensusSame chainEthereum (sequencer orders first)Settlement and data layers
Data availabilitySame chainEthereumSeparate data availability network

Spreading roles across more networks can improve throughput and cost, but security is affected by the weakest layer.

04

Why data availability is moved out

Posting transaction data to Ethereum has been a large share of rollup costs, so designs emerged that store data on networks specialized for it and post only results to Ethereum. Celestia's documentation explains that data availability sampling lets light nodes confirm data has been published without downloading all of it.

Costs fall, but the rollup's security then depends on that data network as well as on Ethereum. If the data network fails, others may be unable to reconstruct the rollup's state, which can make withdrawing assets difficult.

For a hypothetical illustration, if posting a batch's data to Ethereum costs a rollup $0.04 per user transaction and an external data layer costs $0.004, the saving is real. Users are effectively paying less in exchange for relying on a second network's guarantee that the data will stay available when it is needed.

05

Security dependencies

Security in a modular stack works like a chain. A weak proof system in execution, weak verification in settlement, or weak availability guarantees in the data layer weakens the whole. "Secured by Ethereum" therefore needs unpacking: which roles does Ethereum actually perform?

In a hypothetical case, if a rollup executes transactions, settles on Ethereum, and stores data on a separate network whose total stake is a small fraction of Ethereum's, the cost of attacking that data layer can become the ceiling on the system's overall security.

06

Comparing withdrawal paths

For users, the most practical difference is the route out. If a rollup's data is on Ethereum, there may be a path to reconstruct its state from that data and withdraw even if the operator stops.

If the data lives on an external network, the same escape depends on that network continuing to serve the data. Check in the official documentation which design a network uses and whether you could withdraw without the operator's cooperation.

07

Checking each component

Before using a network built on a modular stack, check each layer.

  1. Find out from official documentation which network handles execution, settlement, consensus, and data availability.
  2. Note whether data availability is on Ethereum or an external network, and if external, its size and verification method.
  3. Confirm that the proof system (fraud or validity proofs) is actually live.
  4. See whether users have a path to withdraw on their own if the operator or sequencer stops.
  5. Check who can upgrade the contracts and whether upgrades are delayed.
08

Limits of modular designs

Modular blockchains are evolving quickly, with configurations and terminology that vary between projects and change within months. Splitting roles also creates new risks at the joins between layers.

The security example is hypothetical. When using a real network, rely on Ethereum.org and each project's official documentation along with the date of each, and understand first what happens to your assets if any one layer stops.