Scaling Ethereum with Rollups

Glen Rose
7 min readMay 10, 2022

An introduction to Ethereum, its value and pitfalls, and how teams are working to scale it.

Why is Ethereum so popular?

When Bitcoin was introduced, it allowed for transactions to be posted on the blockchain. Users could create accounts and send money to each other in a permissionless manner. While this was revolutionary, the blockchain’s functionality was limited to transactions only.

Ethereum was the first iteration of this technology that allowed users to upload functional code along with transactions. Contract accounts were created to allow computer programs to create transactions. This opened the door for many use cases. Ethereum introduced the concept of “programmable money”, or the ability to execute arbitrarily complex functions with Turing-complete smart contracts.

After this innovation, developers now had the capability to rethink virtually every type of service offered by the financial industry. From simple escrow accounts and trust entities to complex ideas such as capital raising, collateralized borrowing, product structuring, lending, margin trading, and more. The ability to deploy trustless, interoperable programs created network effects not seen before in financial systems.

Currently, it is the most used blockchain in the world with $103.85B TVL in May 2022 and processed over 1M transactions every day for the last year. Developers and investors from around the world have flocked to Ethereum and have created the majority of the industry’s applications and infrastructure on the EVM standard. Over the last few years, the creation of developer tools such as scaffold-ETH and hardhat, large repositories of source code, and composable assets have made the EVM developer experience the best on the market today, with one exception — the gas fees.

This popularity hasn’t come without pitfalls, and gas fees are the biggest one. The popularity has caused network congestion and driven transaction fees to unsustainable levels. Having a fixed supply of block space naturally causes price increases as demand increases.

Notably, when YugaLabs launched the Otherside in late April, a surge in demand created a closed-bid, first price auction that caused gas prices to skyrocket. Users could expect to pay 2.5+ Ether for a single transaction during this window (over 7000USD at the time) because a large number of users were bidding for a small amount of block space.

Even the flat areas on the chart average out to around 40 dollars per transaction, which make everyday transactions like swaps very inefficient for users. Gas fees are necessary to ensure network security with the current consensus mechanism, however, they are ludicrously high and becoming the largest inhibitor of mass adoption. Transaction fees have reached this level despite Ethereum’s market capitalization being >0.5% of the global equity market. If nothing about the fee structure changes, it will render the network unusable at scale, and Ethereum will fail to become the marketplace for decentralized financial contracts.

Since Ethereum’s conception, users have been hypothesizing about how the network can be scaled without sacrificing its’ key features: security and decentralization.

Scaling solutions

Many teams have come together to solve this issue. Currently, the market for scaling solutions can be segmented binomially by the intended outcome of the project — to improve the Ethereum network or create a better network elsewhere.

Alternate layer 1s like NEAR, Avalanche, or Solana are designed to be a better alternative to the current Ethereum paradigm. They increase throughput by creating new blockchains with consensus mechanisms optimized for throughput — sharding on NEAR while chains like Avlanche and Solana are operated by novel algorithms and more centralized powerful nodes. The ultimate goal of these projects is to offer a better product and poach Ethereum users.

Layer 2 blockchains such as Optimism, zkSync, and Arbitrum are intended to increase the throughput of Ethereum by reducing the computational load of the main chain. By processing transactions on a separate chain and relaying proof back to the main chain, these networks reduce the gas footprint on the L1, making transactions cost less for users. These projects were developed to scale the Ethereum network so developers can continue to utilize the existing infrastructure.

These opposing ideas are well summarized in this African proverb: “If you want to go fast, go alone. If you want to go far, go together”. Alternate L1s are trying to break free from the Ethereum model and make fast progress on their own, while L2s are working to ensure the original network succeeds. Arguments can be made about what approach is better but only time will tell.


Rollups are scaling solutions that fall into the L2 category — they perform transaction execution outside of L1 but post they process multiple transactions on a separate chain, “roll them up” into a single transaction, and relay this back to the main chain, increasing the throughput of the L1 and decreasing transaction costs for users.

Rollups aim to scale the main Ethereum network and keep all development on the EVM standard to continue leveraging and growing its network effects. There are two kinds of rollups: zero-knowledge (ZK) or optimistic (OR) that differ in their security models. ZK uses validity proofs while OR uses fraud proofs. Let’s take a closer look at each.

Zero-Knowledge Proofs

Before understanding how ZK rollups work, you should understand the technology behind them. In short, zero-knowledge proofs utilize cryptography to create a method where one party can prove to another party that a given statement is true without conveying any other information other than the statement is indeed true. Click here for a comprehensive breakdown, but that definition should suffice for this article.

Zero-knowledge proofs have two main applications for blockchain technology: privacy and scalability. Privacy is a huge use case that deserves its own article. For now, check out Zcash and Aztec to learn more about identity protection with zero-knowledge proofs.

Zero-Knowledge Rollups for scalability

For scaling, the rollup is providing the L1 proof that all transactions processed in it are legitimate. A batch of off-chain transactions is sent to a third party who runs the computation and submits a validity proof, called ZK-SNARK (succinct non-interactive argument of knowledge), that all the transactions have been executed and are non-fraudulent.

This is very useful on Ethereum because validating thousands of individual transactions at once on the mainnet would be expensive, if not impossible. Instead, the network only has to process a single ZK-SNARK to verify many transactions. A ZK rollup outsources the computation to generate a validity proof that is submitted to the L1, generating consensus and finality for all transactions in the rollup immediately.


  • Substantially cheaper transactions
  • Near instant finality achieved with validity proof
  • Keeps liquidity on Layer 1 — preserving composability while lowering costs
  • Higher throughput


  • Mathematically complex technology: requires substantial computing power
  • Difficult to prove general purpose EVM execution, better suited for transaction verification
  • Computation outsourced to a single third party

Notable L2 scaling projects: StarkNet, zkSync, Loopring

Optimistic rollups

Unlike ZK rollups, which exert computational power to verify transactions, optimistic rollup validators post data to L1 assuming it’s correct, hence the name optimistic. By skipping the verification process, the network is able to expend less computational power.

To prevent fraudulent transactions from being posted, optimistic rollups have a delay time dispute (DTD) period to verify fraud proofs and detect fraudulent transactions. During this period, network participants can challenge transactions by submitting a fraud-proof, causing the network to execute the transaction again. If re-execution proves the transaction is fraudulent, the validator that submitted the transaction is slashed (loses some of their staked assets). If the fraud proof is incorrect the party who submitted it will be slashed. This mechanism design disincentivizes bad actors from submitting invalid transactions or incorrect fraud proofs.

This less computationally intensive rollup provides a great opportunity for more complex data to be posted to layer 2. ZK rollups are great for processing transaction data, but struggle to handle smart contracts — generalizability is where optimistic rollups shine, but this comes with drawbacks. Since transactions have to enter the dispute resolution period, users have to wait a week to withdraw their funds.


  • Fully EVM equivalent L2 blockchains allow developers to easily deploy existing smart contracts on layer 2.
  • Access to all EVM developer tools (hardhat, scaffold-ETH, etc.)
  • Complicated EVM execution can be run on the layer 2
  • All data is visibly stored on the chain


  • DTD delay of the verification period causes a week-long delay to finality on layer 1
  • Assumptions have to be made that there is at least one honest validator/network participant
  • Less privacy
  • Cannot reach the capital efficiency of ZK without sacrificing their security

Notable L2 scaling projects: Arbitrum, Optimism, and Metis.

Arbitrum is leading the OR market with 80 Defi protocols and 1.67B TVL, Optimism is in second with 38 protocols and $368.56m TVL

Future outlook

Currently, there is a lot of ongoing development in both the optimistic and the ZK rollup camps, alongside a heated debate on who is going to win out. In some cases such as scaling smart contracts, optimistic rollups currently seem better, and in others such as processing large quantities of transactions, ZK outperforms. Investments made by a16z in both technologies signal they believe both will have their respective share of the market.

Many alternatives with low transaction fees will be developed, this won’t be a unique point of differentiation in the future. In my opinion, the biggest driver of success will be the number of valuable products that are built, and this depends on the community of developers. Optimistic rollups currently offer the most developer-friendly experience as existing code can be deployed directly from the EVM to the L2. However, when ZK rollups are able to replicate this EVM equivalence(zkSync and Starkware are close), they will likely win out because of their capital efficiency. If two chains offer the same fees, security, and EVM compatibility, but one requires a week delay to withdraw funds and the other does not, it’s obvious where LPs would rather deposit their funds.