Comparison note · 2026-11-06 · Refs: ethereum.org scaling docs, ETHBEAT (formerly L2BEAT), DefiLlama
The useful question is not which layer is faster. It is where a network gets its security, and what a user can do if the operator stops cooperating. A layer 1 secures itself. A layer 2, in the strict sense, borrows security from a layer 1 and gives users a way to exit to it. Anything that runs its own consensus and only bridges to Ethereum is a separate layer 1 or a sidechain, whatever its marketing says.
Side by side
| Dimension | Layer 1 (e.g. Ethereum mainnet) | Layer 2 rollup |
|---|---|---|
| Security source | Its own validator set and consensus | The layer 1 it settles to, via posted data and proofs |
| Transaction ordering | Validators and block builders | Usually a single sequencer operated by the rollup team |
| Data availability | On the chain itself | Posted to the L1 as calldata or blobs (rollups) or kept off-chain (validiums, optimiums) |
| Exit if operator fails | Not applicable | Forced inclusion or escape hatch to L1, if implemented |
| Withdrawal to L1 | Not applicable | About 7 days for optimistic rollups; after proof acceptance for ZK rollups |
| Upgrade risk | Social consensus and client releases | Upgrade keys, Security Councils, exit windows |
| Fees | Set by L1 block space demand | L2 execution plus the cost of posting data to L1 |
How ethereum.org draws the line
ethereum.org’s scaling overview separates on-chain scaling (changes to layer 1 itself) from off-chain solutions built on top. It describes layer 2 as solutions that “derive their security directly from layer 1 Ethereum consensus,” and sidechains as chains that “derive their security separately from Mainnet.” That one distinction does most of the analytical work:
- Rollups execute transactions off-chain and post the data needed to reconstruct state to Ethereum. If the rollup’s operator disappears, the data is on L1 and users can, in principle, prove their balances and withdraw.
- Validiums use validity proofs like ZK rollups but, per ethereum.org, the “data is not stored on the main layer 1 Ethereum chain.” State correctness is proven, but availability depends on whoever holds the data.
- Sidechains are independent chains with their own validators. A bridge connects them to Ethereum, and the bridge is only as secure as its own verification. See cross-chain bridges.
- State channels let participants transact off-chain through a multisig contract and settle the result on mainnet.
Data availability is the hinge
A rollup that posts its data to Ethereum lets anyone recompute its state. ethereum.org’s optimistic rollups page explains that this is what lets users prove ownership and withdraw even if the operator goes offline. Since Ethereum added blob space for rollup data, most rollups post to blobs, which are cheaper than calldata. The same page notes that blobs are pruned after roughly 18 days, while calldata stays on-chain permanently; long-term access to blob data relies on archives. The 18-day window is long enough for fraud proofs and exits, which is what security requires.
Two proof models
Rollups prove correctness in one of two ways:
- Optimistic rollups assume batches are valid and allow a challenge window, which ethereum.org describes as “roughly seven days,” during which anyone can submit a fraud proof. Withdrawals to L1 wait out that window unless a liquidity provider fronts the funds for a fee.
- ZK rollups post a validity proof with each batch. ethereum.org’s ZK rollups page says there are “no delays when moving funds from a ZK-rollup to Ethereum” once the verifier contract accepts the proof, at the cost of expensive proof generation.
The trade-offs are covered in optimistic vs ZK rollups.
Where the activity is
ETHBEAT, the analytics site formerly called L2BEAT, showed about $37.5 billion in total value secured across tracked projects on 2026-11-06. Its rollup list on the same date had Base at about $14.2 billion, Arbitrum One at $9.4 billion and OP Mainnet at $1.5 billion, all optimistic rollups at Stage 1. For comparison, DefiLlama’s DeFi TVL on Ethereum mainnet was about $43.3 billion on 2026-11-06. The two figures measure different things (value bridged and minted on a rollup versus value locked in DeFi contracts on a chain) and should not be compared directly.
Trust assumptions that remain on most L2s
Being a rollup does not remove operator risk. Three items to check on any specific network:
- Sequencer. Most rollups run one sequencer. ethereum.org lists among the cons that “centralized rollup operators (sequencers) can influence transaction ordering.” The question is whether users can force a transaction through L1 if the sequencer censors them.
- Proof system status. Some rollups run permissioned or no fraud proofs. ETHBEAT’s stages start with whether a project uses “a proper proof system” and whether at least five external actors can submit a fraud proof.
- Upgrade control. Whoever can upgrade the rollup’s L1 contracts can change the rules. ETHBEAT’s Stage 1 requires at least a 7-day exit window for upgrades initiated outside the Security Council, and Stage 2 requires at least 30 days.
ETHBEAT’s own page notes that stages do not reflect project security as a whole; they measure how far a rollup has moved from operator control toward code control.
What this means in practice
For most users, L2s exist because Ethereum mainnet block space is scarce and expensive. The cost of cheaper transactions is a set of trust assumptions about sequencers, upgrade keys and proof systems, plus bridge risk when moving assets between networks. A layer 1 other than Ethereum carries none of the L2 trust assumptions but replaces them with its own validator set’s security, and bridging to it brings back the bridge risk.
This comparison describes architectures; it is not guidance on which network to use or hold assets on. For network-level data, see the layer 2 comparison; for the risk framework, see the DeFi risk checklist.