What Are Layer 2s? Rollups, Fees, and Ethereum's Scaling Problem

The problem Layer 2s actually solve

Ethereum’s base blockchain (often called Layer 1, or L1) can only process a limited number of transactions per block, and every transaction competes for that limited space through gas fees, described in What Crypto Actually Is. When demand for block space rises, whether from a popular NFT mint, a busy DeFi market, or just general activity, fees rise with it, and a simple swap that might cost cents during quiet periods can cost tens of dollars during a busy one. That’s a real, practical limitation, not a temporary bug: Ethereum deliberately prioritizes decentralization and security over raw transaction throughput, which is a defensible tradeoff but leaves a genuine gap for anything that needs to be cheap and fast.

A Layer 2 (L2) is a separate network built on top of Ethereum that handles transactions itself, much faster and cheaper than doing them directly on L1, while still relying on Ethereum for its underlying security. The core idea: do the expensive, security-critical work (final settlement) on Ethereum, and do the cheap, frequent work (actually processing individual transactions) somewhere else, then periodically report back to Ethereum in a compressed, verifiable form.

Rollups: the dominant approach

The most common Layer 2 design is a rollup. A rollup processes transactions off of Ethereum’s main chain, then bundles (“rolls up”) a large batch of them into a single, compact piece of data that gets posted back to Ethereum. Instead of Ethereum verifying every individual transaction in that batch, it just needs to confirm the batch itself is valid, which is far cheaper per transaction than processing each one natively. There are two real approaches to how a rollup proves its batch is valid, and they work quite differently.

Optimistic rollups

An optimistic rollup (used by Arbitrum, Optimism, and Base, among others) assumes transactions in a batch are valid by default and posts them to Ethereum without proving it upfront. To keep that assumption honest, there’s a challenge window, typically around seven days, during which anyone can submit a fraud proof showing a specific transaction in the batch was invalid, which would then roll back the fraudulent part. This design is relatively simple to build and highly compatible with existing Ethereum smart-contract code, which is a major reason Arbitrum, Optimism, and Base (built by Coinbase) became some of the largest and most widely used L2s early on: developers could deploy existing Ethereum applications with little modification.

ZK-rollups

A zero-knowledge rollup (ZK-rollup), used by networks like zkSync Era and Polygon zkEVM, takes a mathematically different approach: instead of assuming validity and allowing a challenge period, it generates a cryptographic proof (a “validity proof”) that mathematically demonstrates the batch is correct before it’s ever posted to Ethereum. Because the proof itself is verified on-chain, there’s no need for a week-long challenge window; a ZK-rollup’s transactions can generally be considered final much faster, often within minutes to under an hour, once the proof is verified. The tradeoff has historically been engineering complexity: generating these proofs efficiently, and making them work smoothly with the exact same code developers already write for Ethereum, has been a harder problem to solve than the optimistic approach, though that gap has been narrowing.

How they actually achieve lower fees while inheriting Ethereum’s security

The fee reduction comes mainly from batching: instead of every individual transaction paying separately to be verified on Ethereum, hundreds or thousands of transactions share the cost of a single, compressed data submission to L1, spreading Ethereum’s base cost across many users at once. The security comes from the fact that the batch’s underlying data (or, for a ZK-rollup, its validity proof) is still ultimately anchored to Ethereum: an L2 can’t simply rewrite history or steal funds without that fraud being detectable and (for an optimistic rollup) provable, or (for a ZK-rollup) mathematically impossible to fake in the first place. This is the meaningful difference between a Layer 2 and a fully separate blockchain: an L2 is deliberately built to inherit Ethereum’s security guarantees rather than establishing its own from scratch.

The honest tradeoffs

• Withdrawal delays on optimistic rollups. Moving funds from an optimistic rollup back to Ethereum’s mainnet natively takes roughly the length of the challenge window, commonly about seven days, because Ethereum needs that window to have passed before treating the withdrawal as final. Third-party “fast withdrawal” services exist that front the funds for a fee, bypassing the wait, but that introduces a separate counterparty you’re trusting instead.
• Bridge risk. Moving assets between Ethereum and an L2, or between two different L2s, generally requires a bridge: a separate smart-contract system with its own code and its own history of exploits, distinct from either chain it connects. Bridge hacks have been among the largest single categories of crypto losses, and that risk exists independent of how secure the L2 or Ethereum itself is.
• Liquidity fragmentation. With activity spread across Ethereum mainnet and multiple L2s (Arbitrum, Optimism, Base, and others, each with their own pools and markets), the same asset can have thinner liquidity, or a different price briefly, on one network versus another, and moving between them takes time and a bridging step rather than being instant.

Frequently asked questions

Do I need a different wallet to use a Layer 2? No, generally the same wallet (like MetaMask) works across Ethereum and its L2s; you typically just add the L2 network’s connection details and bridge or acquire funds on that network. The wallet and seed phrase itself don’t change.

Is a transaction on a Layer 2 as secure as one on Ethereum mainnet directly? It depends on the specific rollup and how much time has passed. A ZK-rollup transaction is generally considered secure quite quickly once its validity proof is verified on Ethereum. An optimistic rollup transaction is technically still within its challenge window until roughly a week passes, though in practice a successful, provable fraud attempt long after a transaction has been widely built upon is rare.

Are Layer 2s only relevant to Ethereum? The rollup concept has been adopted elsewhere, but it originated as, and remains most associated with, a solution to Ethereum’s specific scaling limitations; other blockchains have taken different approaches to the same underlying throughput problem.

How does this relate to DeFi and DEXs? Most major L2s host their own DEXs and DeFi protocols, often the same applications available on Ethereum mainnet, covered generally in What DeFi and a DEX Actually Are. Using DeFi on an L2 works the same way conceptually, just with lower fees and, depending on the L2 type, different withdrawal characteristics back to mainnet.

Risk