Layer 2 and rollups - how Ethereum became cheaper and faster

The Ethereum main network is reliable, but block space is limited, and when there are many users fees rise. The answer is second-layer networks (Layer 2, L2): they do the computation themselves and "borrow" security from Ethereum.

The problem: blocks are not elastic

Every Ethereum node checks all transactions. That makes the network reliable but limits throughput: on the order of tens of transactions per second. If the limit is raised, requirements for nodes grow and only powerful computers can take part. Decentralisation would have to be sacrificed.

The L2 idea: do not expand the main network, move the work out of it. Thousands of transactions are executed on the L2, and only a compressed "digest" and a proof of correctness are sent to Ethereum.

What a rollup is

A rollup is a type of L2 that:

1. accepts users' transactions and executes them itself;
2. publishes data about them to Ethereum so that anyone can restore the state;
3. proves to the main network that the result is correct.

Thanks to step 2 you keep the ability to recover your money even if the L2 operator vanishes: the data is open, and the rules are checked by a contract on Ethereum.

An analogy: Ethereum is the supreme court and a rollup is a magistrate who settles small cases quickly, but every decision can be appealed to the supreme court.

Two kinds of rollups

More on the internals in the article rollups under the hood.

The sequencer

In practice transactions on an L2 are ordered and confirmed by a sequencer - the operator's server. It gives a quick "pre-confirmation" within fractions of a second, which is why you see an instant result. Finality comes later, when the data lands on Ethereum.

The downside of most L2s today: there is one sequencer, and its outage can delay transactions. But there is a "forced inclusion" mechanism through an Ethereum contract that lets you bypass the sequencer.

Fees on L2

An L2 fee consists of two parts:

• the cost of computation on the L2 itself (cheap);
• the cost of publishing data to Ethereum (it dropped sharply after "blobs" arrived in 2024).

In practice a token transfer on an L2 usually costs pennies, whereas on the main network it ranges from fractions of a dollar to tens of dollars under congestion. For services like ours this matters: the lower the fee, the better small and mid-sized deals pay off (how to choose a network).

What a user needs to know

• It is a separate network. It has its own Chain ID (Base - 8453), its own RPC and its own balances. The address is the same, but the money is on a different "floor".
• Tokens are different. USDC on Base and USDC on Ethereum are different records in different contracts. Moving between networks goes through a bridge.
• Withdrawing to Ethereum is slow (for optimistic rollups). Third-party services offer fast withdrawal - for a fee.
• Gas is paid in ETH (on Base and other Ethereum-based L2s). You need to keep a little ETH on the right network.

L2 risks

• Sequencer centralisation and administrative keys that can upgrade contracts.
• Bugs in the rollup's and bridge's code.
• Dependence on the operator when withdrawing funds.

That is why it matters to look at which stage of decentralisation a network is at.

Example: the same deal on two networks

Suppose you need a deal for 100 USDC. On the main network creating the offer, accepting and confirming are three transactions, and with average congestion that can total 10-30 dollars. On Base the same three actions cost cents in total. For a 100 USDC deal the choice is obvious. For a 50,000 USDC deal a 20-dollar difference is immaterial and many will prefer the main network.

What to read next

Compare the networks the service works on: how to choose a network. For those who want to look deeper - rollups under the hood.