Layer 2 and Rollups: How Ethereum Learned to Scale

Every blockchain faces the same tension. The more transactions a chain processes, the more powerful the computers needed to keep up, and the fewer people can run a node and verify the network themselves. Bitcoin and Ethereum have both chosen to keep their base layers relatively lean so that verification stays widely accessible.

For Ethereum, that choice had a visible cost. In busy periods, block space became scarce, and fees for a simple transfer or a token swap could rise so far that small transactions made little economic sense. The question was how to grow without giving up what made the network valuable in the first place.

The answer that emerged over several years is a layered model. Ethereum serves as a secure foundation, while much of the everyday activity moves to Layer 2 networks, above all to so-called rollups. This article explains how they work, what changed in 2024 and where the open questions lie.

The scaling problem of a base chain

A blockchain is a shared ledger that thousands of independent computers keep in sync. Each of them, called a node, checks every transaction against the rules. This redundancy is what makes the system hard to manipulate, but it also means that the network can only be as fast as the slowest node it wants to include.

Larger blocks or shorter block times would increase capacity, but they would also raise the requirements for hardware and bandwidth. Over time, fewer participants could afford to verify the chain independently. Ethereum's developers concluded early that this trade was not worth it and looked for ways to add capacity on top instead.

The term Layer 1 refers to the base chain itself. A Layer 2 is a separate system that processes transactions on its own but anchors its security to Layer 1, so that users do not have to trust the Layer 2 operator blindly.

How rollups bundle transactions

A rollup collects many transactions, executes them outside Ethereum's main chain and then publishes a compressed summary back to it. Two things are posted: the transaction data, so that anyone can reconstruct the state, and a commitment to the result, meaning the new balances after all transactions have been applied.

Because many transactions share the cost of a single posting to Ethereum, each one becomes much cheaper. And because the data sits on Ethereum, users can in principle recover their funds even if the rollup operator disappears. This property, known as data availability, is what separates a rollup from an independent chain with its own rules.

Rollups come in two main families. They differ in how they convince Ethereum that their results are correct.

Optimistic rollups

Optimistic rollups assume that posted results are valid unless someone proves otherwise. After each batch, a challenge window opens, in many designs about seven days, during which anyone watching can submit a fraud proof. If a result turns out to be wrong, it is reversed and the dishonest party is penalized.

The model is relatively simple and compatible with existing Ethereum software. Its main drawback is the waiting time for withdrawals back to Ethereum, which have to outlast the challenge window unless a third party offers a faster exit for a fee.

Zero-knowledge rollups

Zero-knowledge rollups, often shortened to ZK rollups, take the opposite approach. With every batch, they submit a cryptographic validity proof that Ethereum can check quickly. If the proof verifies, the result is correct by mathematics, not merely accepted because nobody complained.

The name is slightly misleading, because the key feature is not secrecy but a compact proof of correct computation. Generating these proofs is demanding and the technology is younger. In return, withdrawals can become final much sooner once a proof has been accepted.

Dencun and blobs: why fees fell in 2024

For years, rollups had a cost problem of their own. They wrote their data into the same space that regular Ethereum transactions used and competed for it at the same price. Data made up the largest part of what users paid on Layer 2.

The Dencun upgrade in March 2024 changed that with EIP-4844, also known as proto-danksharding. It introduced blobs, a separate type of data space attached to blocks. Blobs have their own fee market and are deleted by nodes after a limited period, which is long enough for anyone to check and copy the data but avoids storing it forever.

Fees on Layer 2 networks dropped significantly afterward. Blobs are also a step in a larger plan, in which Ethereum gradually increases how much rollup data it can carry without forcing every node to download everything.

Ethereum did not try to become the fastest chain. It chose to become the most reliable place where faster systems settle their accounts.

The trade-offs: sequencers, bridges, fragmentation

Rollups move a lot of work away from Ethereum, and with it some of the guarantees. The open issues are well known among developers and worth understanding before you move funds.

• Sequencers. Most rollups currently rely on a single sequencer, a server run by the project team that puts transactions in order. If the rollup is built correctly, it cannot steal funds, but it can delay or hold back transactions, and an outage can halt the network for a while.
• Upgrade keys. Many rollups can still be changed by a small group holding administrative keys. These keys are meant as a safety net during the early phase, but they mean that users depend on the integrity of that group.
• Bridges. Moving assets between chains requires bridges, and several of the largest thefts in crypto history have targeted them. Official rollup bridges inherit Ethereum's security more directly than third-party bridges between unrelated chains.
• Fragmentation. Each rollup has its own balances, liquidity and applications. Assets on one rollup are not automatically available on another, which splits users and capital across many networks.

Projects are working on all of these points, from shared or decentralized sequencers to standards for moving assets between rollups. Independent trackers rate rollups by how far they have progressed in removing their "training wheels", meaning the operators' ability to intervene. That rating is a useful sign of maturity.

What users notice in daily use

For most people, the difference shows up in the wallet. Transactions on a rollup typically cost far less than on Ethereum itself, and confirmations appear within seconds. Many applications now run on several rollups at once, and some trading platforms allow deposits and withdrawals directly through Layer 2 networks.

The flip side is a new kind of attention. You need to check which network your wallet is connected to and which network a recipient expects. Because addresses often look identical across Ethereum-compatible chains, sending to the right address on the wrong network is a common mistake, and recovering such funds can be difficult or impossible.

Withdrawals deserve particular care. Leaving an optimistic rollup through its official bridge can take about a week, and faster third-party routes add costs and another layer of trust. Planning ahead avoids unpleasant surprises.

Where scaling goes from here

Ethereum's roadmap continues in the direction set by Dencun. Further upgrades aim to increase the number of blobs per block and to let nodes verify data availability without downloading all the data. At the same time, rollups are working to decentralize their sequencers and to reduce their dependence on administrative keys.

Whether the layered model will one day feel as seamless as a single network is still open. Today, it already handles a large share of the activity in the Ethereum ecosystem at far lower cost than the base chain could. For users, the practical lesson is simple: Layer 2 is a convenience with its own rules, and those rules are worth knowing.