Layer 2 separates execution from settlement
A Layer 1 is a base blockchain with its own rules for agreeing on transactions and ledger state. A Layer 2 moves much of the everyday processing into another system while relying on the base chain to enforce important guarantees. In an Ethereum rollup, transactions execute on the rollup, while Ethereum provides the foundation for data publication and settlement.
This division can make transactions cheaper because many users share the cost of using scarce Layer 1 resources. It also introduces sequencers, bridge contracts, proof systems, and additional recovery paths. “Layer 2” describes a relationship between systems; it does not by itself establish how safe any particular network is.
This article focuses on Ethereum-style rollups. Other scaling designs divide responsibility differently, so the same explanation should not be applied automatically to every network using the label.
Why increasing Layer 1 capacity has a cost
A blockchain is not simply one operator’s database. Independent participants need to verify that transactions follow its rules. Larger blocks and more computation can increase capacity, but can also demand more bandwidth, storage, and processing power from the computers doing that verification.
If participation becomes too demanding, fewer people can check the system independently. Layer 1 upgrades are possible, but capacity cannot be increased without considering those costs.
Rollups take another route. They execute transactions outside Ethereum’s main execution environment, group activity into batches, and publish the necessary information back to Ethereum. The work is distributed across layers rather than making Ethereum directly execute every action.
A rollup submits commitments to its updated state: balances and application information. A commitment is a compact cryptographic representation, not a readable list of all accounts. It also cannot establish correctness by itself. A fraudulent collection of balances can have a perfectly valid cryptographic fingerprint.
Follow a payment across the layers
Suppose a user starts with ETH on Ethereum and deposits through a rollup’s native bridge. An Ethereum transaction records the deposit and locks the ETH under the bridge’s rules. After processing that message, the rollup credits the corresponding Layer 2 balance.
Changing the selected network in a wallet does not perform this deposit. It changes which network the wallet reads and sends transactions to. The same address can have a different balance on each network because the ledgers are separate.
The user now signs a Layer 2 payment. In many rollups, a sequencer orders transactions, after which the rollup executes the payment and updates the balances. The recipient can receive a quick response without every stage of settlement being complete.
Keep three milestones separate:
| Milestone | What it tells the user |
|---|---|
| Sequencer acknowledgement | The ordering service has responded to the transaction |
| Required data in finalized Ethereum blocks | The Layer 1 publication step has reached that chain’s finality |
| Completed native withdrawal | The exit has satisfied the rollup’s settlement rules |
A quick acknowledgement is useful feedback, but it does not establish that the later milestones have happened.
Settlement does not require every payment to return to Ethereum. Users can keep transacting on Layer 2. It establishes the result that Layer 1 contracts can act on, including when funds leave through the native bridge.
Optimistic rollups allow a result to be challenged
An optimistic rollup proposes a result that can stand unless a successful challenge establishes that it is wrong within a defined period. Observers can reconstruct the transactions and use the fault-proof process to dispute invalid claims.
This system relies on more than assuming an operator is honest. The necessary data must be available, honest challengers must be able to detect a fault, and their challenge must reach the contracts enforcing the rules.
The challenge period explains why a native withdrawal can take days even when an ordinary Layer 2 payment appears quickly. Releasing funds immediately would remove the time needed to contest the claim authorizing the exit. Exact timings depend on the particular network’s rules.
The optimistic-rollup documentation describes these mechanisms. For a concrete implementation, the companion article on Base follows how an Ethereum rollup combines sequencing, batching, and fault proofs.
Validity proofs offer a different enforcement route
A zero-knowledge, or ZK, rollup supplies a cryptographic proof that a proposed state update obeyed the rules. Ethereum can verify that proof without repeating all of the original computation. Generating the proof itself can require substantial computation and specialized software.
This removes the need for the same fraud-challenge window, but does not guarantee instant withdrawals. Proof generation, publication, and Layer 1 finality still take time.
The name also creates a common misunderstanding: a ZK rollup is not automatically private. Proving correct execution and hiding transaction details are separate capabilities. A system can use this proof technology while publishing transaction information that remains visible to others.
Neither proof approach makes its contracts immune to bugs. It defines how correctness is enforced under the design’s assumptions, rather than promising that every implementation and application works perfectly.
Data availability lets others check the ledger
A claim about final balances is insufficient if the operator withholds the data needed to reconstruct them. Other participants need that information to verify the state and follow the system’s recovery procedures independently.
Ethereum rollups publish required data using mechanisms such as calldata or blobs. Blobs provide temporary availability rather than permanent retention by every Ethereum node. Preserving access to historical data involves additional storage outside that temporary window.
This is why “off-chain execution” does not mean the transaction exists only in someone’s wallet. Execution occurs on the rollup; the base chain hosts the settlement contracts and the published information connecting that execution to Layer 1.
A sidechain has a different relationship. It uses its own consensus and security assumptions. A bridge can connect it to Ethereum without Ethereum enforcing the correctness of its entire ledger. Cheap fees and familiar applications do not tell you which of these architectures you are using.
Fast bridges add another set of conditions
A native bridge follows the rollup’s settlement path. A faster service may instead use liquidity providers to advance funds on the destination network and settle later. That can reduce the user’s wait, but adds fees, liquidity requirements, and different trust assumptions.
The underlying settlement delay has not disappeared. Someone else has taken on the interval between paying the user and completing the slower route.
Operators and upgrades matter too. Sequencer outages can interrupt normal use, and censorship can delay inclusion. Some rollups support a Layer 1 route to force transactions through, but that fallback can be slower and more expensive. If a small group can upgrade crucial contracts, its authority is part of the system’s security model.
Judge the responsibilities behind the layer number
Lower fees can make small payments and frequent application interactions practical. The total cost still includes execution, data publication, and any bridge transactions needed to enter or leave. A rollup may use ETH for fees; the existence of a branded network does not automatically require a separate token.
The useful questions are concrete: where does execution happen, who can order or delay transactions, how can an incorrect result be rejected, and how can a user exit if the operator stops cooperating? Those answers explain the tradeoff. Layer 2 extends what a base chain can support by dividing the work, while asking users and developers to understand a more complicated route from transaction to settlement.
