A rollup is a layer 2 network that executes transactions off Ethereum, bundles them, and posts the data back to Ethereum. This article explains how that works, how optimistic and zero-knowledge rollups differ, and why every rollup fee includes a cost for Ethereum.
For the bigger picture of layers, see layer 1, layer 2 and layer 3 explained.
How a rollup batches transactions
When you send a transaction on a rollup such as Base, it does not go straight to Ethereum. It goes to the rollup's operator, which executes it and updates the rollup's own state, such as balances. A few seconds later, many such transactions are compressed into a batch. The batch is then posted to Ethereum in one transaction.
The word "roll up" describes this: many transactions are rolled into one. The cost of the Ethereum transaction is shared by everyone in the batch, which is the main reason fees are lower than on Ethereum itself.
The sequencer
The sequencer is the server that receives transactions on a rollup, puts them in order, executes them and produces L2 blocks. It also gives users quick confirmations, often in well under a second, before anything reaches Ethereum.
Most large rollups today run a single sequencer operated by the team behind the chain. This is simple and fast. It also brings trade-offs, described neutrally here:
- Ordering power. The sequencer chooses the order of transactions, so in theory it could delay or reorder them.
- Downtime. If the sequencer stops, normal use of the rollup stops. Rollups have a fallback: users can submit transactions through a contract on Ethereum, though this is slower and more expensive.
- Cannot steal funds on its own. The sequencer cannot create a valid withdrawal for money it does not own, because the proofs or checks on Ethereum would reject it.
Teams across the industry talk about decentralizing sequencers over time. The current status differs per chain, so check each project's documentation.
Posting data to Ethereum: calldata and blobs
A rollup must publish its data on Ethereum so anyone can verify the chain. Historically it did this in calldata, a field of an ordinary Ethereum transaction. Calldata is permanent and competes with all other Ethereum activity, so it was expensive.
In 2024 Ethereum added blobs through EIP-4844. A blob is a large chunk of data that Ethereum nodes keep for a short time, not forever, and it has its own fee market separate from normal gas. This is much cheaper for rollups, and it lowered the typical cost of L2 transactions. You can read the specification at eips.ethereum.org.
Optimistic rollups
An optimistic rollup assumes the batches posted are correct unless someone proves otherwise. That is the optimism in the name.
- After a batch is posted, there is a challenge window. During it, anyone who spots an invalid result can submit a fraud proof to Ethereum.
- If the fraud proof is valid, the wrong result is discarded.
- If nobody challenges, the result is final on Ethereum after the window closes.
The challenge window is about 7 days for withdrawals to Ethereum through the official bridge. Everyday use on the rollup is not delayed. Only moving funds back to L1 through the canonical bridge has to wait. Fast third-party bridges can speed this up for a fee, and bridges and their risks explains how.
Three chains on ChainMonitor are optimistic rollups: Arbitrum One, Base and OP Mainnet. Base and OP Mainnet use the OP Stack. Arbitrum uses its Nitro stack. See the guides for Arbitrum, Base and Optimism.
Zero-knowledge rollups
A zk rollup (zero-knowledge rollup) does not rely on a challenge window. Instead, each batch comes with a validity proof, a mathematical proof that the new state follows from the old state under the rules. Ethereum checks the proof with a smart contract. If it is valid, the batch is accepted.
The benefits are quicker finality on L1, because no long waiting period is needed, and a strong guarantee that invalid state cannot be accepted. The costs are that generating proofs needs heavy computation, and the technology is more complex to build, especially when supporting all Ethereum features. This area is developing quickly, and several zk rollups exist. None of the seven chains on ChainMonitor is a zk rollup.
| Feature | Optimistic | ZK |
|---|---|---|
| Basic assumption | Correct unless challenged | Proven correct |
| How invalid batches are stopped | A fraud proof during the challenge window | A validity proof with every batch |
| Withdrawal to L1 | About 7 days (canonical bridge) | Faster, after proof is verified |
| Computation load | Lower | Higher for proofs |
| On ChainMonitor | Arbitrum, Base, OP Mainnet | None |
The L1 data fee in every L2 fee
An L2 transaction fee has two parts:
- L2 execution fee. Pays for running your transaction on the rollup. It is usually tiny.
- L1 data fee. Pays for your share of posting data to Ethereum. This is often the larger part of the fee.
The L1 data fee changes when Ethereum data costs change, so the total fee on an L2 can move even when the L2 itself is quiet. On the gas tracker, the numbers for L2s include this L1 data fee, so you see the true cost. To dig deeper, read gas and fees.
What this means for you
- For daily use, optimistic and zk rollups feel similar: fast and cheap.
- If you plan to move funds back to Ethereum, check how long it takes. For the optimistic ones, plan for about 7 days or use a bridge and read its risks.
- Your funds are held by contracts on Ethereum, but a rollup still has its own code, its sequencer and its upgrade rules. Those are separate risks.
Frequently asked questions
What is a rollup in crypto?
A rollup is a layer 2 network that processes transactions off Ethereum, bundles them into batches, and posts the data to Ethereum so that it can inherit Ethereum's security.
Why does withdrawing from Arbitrum, Base or Optimism take 7 days?
These are optimistic rollups. After a withdrawal is requested through the official bridge, there is a challenge window of about 7 days during which anyone could prove it is invalid. Funds are released after the window ends.
What is the difference between optimistic and zk rollups?
Optimistic rollups assume results are correct and allow challenges for a period. ZK rollups attach a mathematical validity proof to every batch, so no waiting period for challenges is needed.
What is a sequencer?
A sequencer is the server that orders and executes transactions on a rollup and gives fast confirmations. Most large rollups use one sequencer run by the project team, which is fast but is a point of centralization.
What are blobs on Ethereum?
Blobs are cheap, temporary data attachments introduced by EIP-4844. Rollups use them to post their data to Ethereum at a lower cost than calldata.