Layer Two Rollups

What Rollups Move Off-Chain

At their core, rollups are a layer-2 scaling approach that shifts computation and state storage off the Ethereum mainnet. While transactions are executed offchain, critically, all transaction data is posted back to the base chain. This ensures Ethereum remains the ultimate arbiter of state and settlement, while the heavy lifting of execution happens elsewhere.

The two most common forms of rollups are optimistic and zero-knowledge. But both models rely on data availability from the base layer, which provides the final security guarantees. Rollups post data and proofs to a bridge contract on Ethereum, which holds user assets. However, this makes the bridge a risk concentration point.

Today, the largest ongoing cost in rollups is the cost of publishing that escalated data on the base chain - and as transaction volumes scale, those costs can become prohibitive.

How Optimistic Rollups Work

The simplest way to understand optimistic rollups is this: they assume the posted results are valid, unless proven otherwise. After executing transactions offchain, the rollup posts a summary batch to the Ethereum network. There is then a short "challenge window" during which anyone can submit a fraud proof if they believe the results are incorrect.

If no fraud is found within this window, the state change is finalized on Ethereum. However, this process traditionally includes a withdrawal delay of about 7 days on optimistic systems. That is by design, not an oversight. The withdrawal delay acts as a final layer of security and provides time for fraud proofs to be submitted.

How Zero-Knowledge Rollups Differ

In contrast, zero-knowledge rollups do not rely on external validity assumptions. Instead, they bundle transactions into batches and cryptographically prove the validity of each batch. This zk proof is then posted to the base chain, where Ethereum's consensus layer verifies the proof.

Because zero-knowledge proofs affirm the correctness of offchain execution upfront, zk rollups do not need a challenge period. Upon receiving the valid proof, the state change is immediately finalized - meaning faster withdrawals compared to optimistic rollups.

However, the tradeoff is that generating these cryptographic proofs requires more computational work than an optimistic assumption. So zero-knowledge rollups tend to have higher upfront proving costs.

Where Rollup Costs Concentrate

Whether optimistic or zero-knowledge, both rollup types aim to minimize the amount of data required to be posted to the Ethereum base layer. But as a core component of the model, the cost of publishing that offloaded data remains a persistent expense that scales with usage.

Additionally, with rollups, security rests critically on the bridge contract used to hold rolled-up assets and facilitate cross-chain movement. The bridge becomes a concentrated attack vector, as demonstrated in some incidents involving bridge hacks.

What "Not Rollups" Are

It is important for users and developers to understand that not all offchain scaling solutions are rollups. Systems that post only a hash and keep the data offchain in a different location are a different trust model entirely, and are not what this guide defines as rollups.

Vitalik Buterin makes this distinction clear in his comprehensive rollup guide, which explains how data availability from the base chain — specifically, data availability of the input data — is a defining characteristic of the rollup model.

The Rollup Tradeoff: A Final Note

Rollups presently strike a balance between offloading execution and maintaining onchain security. They are not the only way to scale, but they offer a path that still keeps Ethereum at the core.

The bridge contract and the data published onchain are key ingredients in the rollup formula, but also present risks and a major cost driver. A pending question is whether rollup systems can evolve to further reduce data written to Ethereum and the reliance on offchain components - even as they scale to serve transactions.