Layer 2 rollups explained: how Ethereum scales without bigger blocks

What is a layer 2?#

Layer 1 is Ethereum Mainnet, the base blockchain described in what is Ethereum. Layer 2 is a collective term for systems that handle transactions off Mainnet while taking advantage of Mainnet’s decentralized security ^[1]. ethereum.org draws a careful line: a true layer 2 derives its security from Ethereum. Sidechains, validiums and plasma chains also connect to Ethereum but secure themselves in other ways ^[1].

Rollups are the main kind of layer 2. They execute transactions outside layer 1, then post the transaction data to layer 1, where consensus is reached. Because the data sits in Ethereum blocks, the rollup can be secured by Ethereum itself ^[1].

Why can’t Ethereum just make its own blocks bigger?#

Because every full node must process every block. Bigger, faster blocks would need larger nodes and more specialised hardware, which would reduce the number of people who can run one and undermine decentralization ^[3]. Ethereum caps block size for exactly this reason: if blocks could grow without limit, less powerful nodes would fall behind ^[4]. The chosen path is to keep layer 1 as a secure settlement layer and move most activity to rollups ^[5].

How does a rollup work?#

From your transaction to Ethereum

Each rollup is controlled by smart contracts on Ethereum that store its blocks, track deposits and record its latest state root, a short fingerprint of every balance and contract on the rollup ^[6]. The sequencer is the operator that orders transactions and submits batches. Because the data is published on Ethereum, anyone can rebuild the rollup’s state; if an operator goes offline, another node can use that data to continue, and users can prove what they own to withdraw ^[6].

What is the difference between optimistic and ZK rollups?#

They differ in how Ethereum becomes sure that a batch was computed correctly.

Summarised from the pros-and-cons tables on ethereum.org. Both types publish enough data on Ethereum for anyone to rebuild the rollup’s state.

In an optimistic rollup, a challenger who spots a wrong result starts a fraud proof: the two sides narrow the dispute down to a single computation step, which a contract on Ethereum then checks. The loser forfeits a bond ^[6]. ZK rollups use proofs called ZK-SNARKs or ZK-STARKs. Some SNARK systems depend on a one-time “trusted setup” ceremony; STARKs avoid that but produce larger proofs, which cost more to verify on Ethereum ^[2]. Despite the name, the proofs are used for compactness, not privacy: a ZK rollup’s data is still public ^[2].

Why are rollups cheaper than Ethereum Mainnet?#

Two reasons. First, batching: the fixed cost of writing to Ethereum is shared by every transaction in a batch ^[6]. Second, compression: a rollup can describe a simple transfer in far fewer bytes, for example by referring to accounts by a short index instead of a full address ^[2].

Bytes needed for a simple ETH transfer

ethereum.org calls this a “fairly optimistic” estimate, since rollup data never fills a whole block, and notes that real implementations offered up to about 2,000 transactions per second when it was written ^[6].

When the EIP-4844 proposal was written, its authors reported that optimistic rollups such as Optimism and Arbitrum often charged fees about 3–8 times lower than Ethereum itself, and ZK rollups about 40–100 times lower ^[7]. Those ratios move with demand on both layers, so treat them as a snapshot, not a price list.

What are blobs, and what did EIP-4844 change?#

Rollups used to post their batches as calldata, ordinary transaction data that stays in Ethereum’s history permanently. EIP-4844, part of the Dencun upgrade, added blob-carrying transactions ^[8]. The Ethereum Foundation scheduled Dencun to activate on mainnet on 13 March 2024 at 13:55 UTC ^[9]. A blob is a large chunk of data that Ethereum Virtual Machine execution cannot access ^[7], priced in its own fee market and usually cheaper than calldata ^[2]. The network only has to keep serving blob data for about 18 days; it is not permanent storage ^[2].

The EIP described this as a reduced cap compared with full data sharding and as temporary scaling relief on the path to sharding ^[7]. Those launch limits have already changed once: the Pectra upgrade, scheduled for 7 May 2025, raised the average from 3 to 6 blobs per block and the maximum from 6 to 9 ^[12]; EIP-7691 sets those as the new target and maximum ^[10]. In the same announcement, the Ethereum Foundation said blobs had cut the layer-1 fees paid by layer 2s by 10–100 times since going live ^[12]. The Fusaka upgrade in December 2025 introduced PeerDAS, a more efficient way for layer 2s to post and retrieve data on Ethereum ^[3]. The Ethereum Foundation scheduled Fusaka for 3 December 2025 ^[11]. Its announcement said layer 2 usage was often reaching the 9-blob limit at the time ^[11]. With PeerDAS, nodes check that blob data is available by sampling pieces of it instead of downloading every blob ^[11]. For how the blob fee mirrors the main fee market, see EIP-1559 explained.

How many blobs can a block hold after Fusaka?#

After PeerDAS, the plan is for Ethereum to raise blob capacity through small upgrades called Blob Parameter Only (BPO) forks instead of waiting for a large named upgrade ^[11]. A BPO fork changes only three settings: the blob target, the blob maximum and how fast the blob fee adjusts ^[13]. It needs no new client code; the new values are set in configuration and take effect at a planned time ^[13]. The Fusaka announcement scheduled BPO1 for 9 December 2025, raising the target from 6 to 10 blobs per block and the maximum from 9 to 15 ^[11]. It scheduled BPO2 for 7 January 2026, raising the target to 14 and the maximum to 21 ^[11]. These are the most recent blob limits in our sources, dated to that November 2025 schedule ^[11]. Later BPO forks can change them again, so treat any blob count as a dated value and check a live source before relying on it.

Are layer 2s as safe as Ethereum?#

Not automatically. ethereum.org says their safety depends on the underlying technology, the security of their smart contracts and how mature the network is, and advises starting with small transactions ^[3]. Common weak points include:

• Centralised sequencers. Many rollups have a single operator that can influence transaction ordering, or censor users by going offline ^[6]. Good designs let users force a transaction through Ethereum instead.
• Upgradeable contracts. A rollup’s contracts on Ethereum can often be changed by a council or team. The independent tracker L2BEAT rates rollups in “stages” from 0 (“full training wheels”) upwards, based partly on whether users can exit if operators misbehave — and warns that stages do not reflect overall project security ^[14].
• Proof systems. An optimistic rollup is only as safe as the assumption that at least one honest node is watching; a ZK rollup depends on correct proof code and, for some systems, a trusted setup ^[2].

How do you move funds to and from a layer 2?#

1. Deposit through the bridge

   You lock ETH or tokens in the rollup’s bridge contract on Ethereum, and an equal amount is created for you on the rollup ^[6].

2. Use the rollup

   Transactions on the rollup pay the rollup’s own fees, which cover its operator and the cost of posting data to Ethereum.

3. Withdraw

   Your assets are burned on the rollup and released on Ethereum once the batch is final — after the roughly seven-day challenge period for optimistic rollups ^[6].

4. Or pay for speed

   Liquidity providers can pay you on Ethereum straight away in exchange for a fee, and wait out the delay themselves ^[6].

What mistakes do beginners make here?#

• Assuming every “layer 2” is a rollup

  Some networks keep their data off Ethereum or use their own validators. ethereum.org treats only systems that derive security from Ethereum as layer 2.

• Expecting instant withdrawals from an optimistic rollup

  The native route back to Ethereum waits for the challenge period, about a week. Fast exits use a third party and cost a fee.

• Sending funds on the wrong network

  The same address can exist on Ethereum and on many rollups. Check which network your wallet and the recipient are using before you send.

• Trusting an unfamiliar bridge or a copycat token

  Bridges can be hacked, and look-alike tokens exist on many networks. Use well-known routes and check token contracts before you send.

• Ignoring who can upgrade the rollup

  If a small group can change the rollup’s contracts at short notice, your funds depend on that group as well as on Ethereum.

Frequently asked questions#

The bottom line#

Rollups are how Ethereum handles more users without asking every node to do more work: execute elsewhere, compress, post the data to Ethereum, and settle there with either a challenge window or a proof. Blobs made that data cheaper to post.

The savings come with extra layers of trust — sequencers, upgrade keys and bridges — that vary from one rollup to the next. Check them before moving meaningful amounts. For how fees work on Ethereum itself, read Ethereum gas fees explained.

Sources#