It’s done, welcome to the post-Ethereum Merge world! After almost half a decade of delays due to technical difficulties, the Ethereum blockchain moved from a proof of work (PoW) to a proof of stake (PoS) consensus mechanism without a hitch on Thursday, 15 September 2022.

The highly anticipated event drastically reduces Ethereum’s carbon footprint, cutting energy consumption by ~99.5% and allowing users to stake their Ethereum (ETH) to secure the network, earning a yield in the process. For those interested in the details of the Merge, we have covered this in our previous blog.

According to Vitalik Buterin, co-founder of Ethereum, “the Merge just symbolizes the difference between early stage Ethereum, and the Ethereum we’ve always wanted”.

However, some problems remain. Namely, the Merge won’t facilitate Ethereum scaling and thus will not reduce the high gas fees.

Yet the Ethereum community has set an ambitious goal of becoming the settlement layer of the emerging Web3 ecosystem, securing hundreds of trillions of dollars in decentralized financial applications and replacing paper contracts and trusted intermediaries with automated, autonomous smart contracts.

Consequently, there is a need for scaling solutions that will allow Ethereum to achieve these ambitious goals. This blog discusses how the Ethereum developer team plans to address this challenge and how other parts of the Ethereum ecosystem, i.e. rollups or Layer 2 protocols, are already scaling the network.

Ethereum’s history

One could say that Ethereum’s scalability issues stem from its success. When Ethereum introduced smart contracts in 2015, it became fertile ground for innovative decentralized applications (dApps), from games and collectibles to financial applications.

The Solidity programming language and the ERC20 standard for creating new programmable tokens proved to be excellent tools for this, and Ethereum became a magnet for the most talented developers in the space.

In December 2017, the non-fungible-token (NFT) game CryptoKitties infamously congested the network with a six-fold increase in network requests, and many feared the network would grind to a halt.

2017 was also the year that Maker, the project behind the decentralized stablecoin DAI, was launched. But it wasn’t until the summer of 2020 that the dramatically expanding decentralized finance (DeFi) ecosystem caused gas prices to spike.

Resign gas fees and competition

Gas fees are caused by activity on the Ethereum blockchain, such as a simple transaction, a trade or an NFT-mint. These fees measure the amount of computational effort required to execute each operation.

The total costs of such transactions in USD results from the following factors:

A simple transfer from wallet to wallet typically costs 21,000 gas, while a trade on Uniswap could cost around 100,000 gas, depending on the complexity of the trade.

When gas prices skyrocketed to an average of 538 gwei on 17 September 2020 and ETH was trading at around $370, for example, users would have to accept gas costs of around $199 per transaction for a trade on Uniswap.

As a result, many users began looking for emerging alternatives. By this point, many Ethereum competitors with higher throughput had emerged, looking to lure users away from Ethereum with lower gas fees. Consequently, Ethereum lost significant market share to its competitors. To date, however, it has managed to defend the magic 50% mark.

Ethereum’s continued popularity is due to its wide range of applications and strong security properties. However, to defend its position as the largest smart contract platform in the long term, Ethereum needs to scale drastically.

Scaling Ethereum means increasing the transaction speed of the network (faster finality) and, more importantly, its throughput (more transactions per second) without increasing the costs of validating them.

Increasing the scalability of a blockchain while preserving its decentralization and security is a challenging task. Vitalik once described this phenomenon as the “scalability trilemma”.

The “scalability trilemma”

The scalability trilemma states that a blockchain generally tries to achieve three properties but can only achieve two with a typical, monolithic blockchain design.

These properties are: scalability, decentralization (the number of nodes) and security.

A monolithic blockchain is one where any change to one component would require an update to the entire network. As such, they are limited by what a single node can handle. Achieving higher throughput and scaling the network is only possible with the help of upgraded hardware for the nodes validating, executing and storing transactions. This, however, ultimately decreases the decentralization of the network, as the costs of running a node increase.

Solana is a good example of this. To achieve the highest throughput, the blockchain often requires sophisticated hardware, meaning node operations are restricted to data centers. Pushing this design to its limits will ultimately put Solana’s governance in the hands of those who can afford the increased hardware requirements to run these nodes.

Ethereum’s off-chain vs. on-chain scaling solutions

To escape the scalability trilemma, the Ethereum community is simultaneously working on off-chain and on-chain scaling strategies that will facilitate increasing the throughput to a much higher level than a single scaling strategy could achieve on its own.

Together, these scaling strategies are referred to as a modular blockchain. Modular blockchain designs are not constrained by the scalability trilemma because they divide the work among Layer 1 and additional Layer 2 solutions.

Ethereum’s roadmap foresees that Layer 1 remains highly secure and decentralized, while Layer 2 solutions focus on scalability, relying on Layer 1 for security. Currently, the most well-adapted Layer 2s of the Ethereum ecosystem are Arbitrum and Optimism.

Key properties of blockchain scaling

Consensus layer

The consensus mechanism ensures that new entries to the blockchain are true and honest and guarantees uptime (the percentage of time during which the system remains in operation), censorship resistance, and security of the network.

The consensus mechanism also enables trust minimization, a unique security property that allows for the replacement of paper contracts or trusted intermediaries.

With Ethereum’s PoS mechanism, the staked amount of ETH tokens is effectively the security budget of the network. In this context, security mainly describes how nodes achieve an honest majority in the face of a certain threshold of malicious actors.

This security is not limited to transactions on the main chain (Layer 1). It also extends to Layer 2 scaling solutions such as rollups, which focus on the execution of new transactions in a more efficient manner.

Execution layer

The execution layer is where the computations needed to update the state of the blockchain take place. Transactions and updates of contracts are executed here.

These can take place on a Layer 1 blockchain, incurring high gas costs as a result of limited blockspace, or be handled by a Layer 2 solution, which can perform multiple transactions at significantly lower gas costs.

Data (Availability) layer

When a block is updated, the data is stored on all computers running full nodes of the blockchain. It is then available for anyone to see on the public ledger.

This data is very limited and higher activity on the blockchain increases the demand for data on Layer 1. As a result, gas fees increase. This is where off-chain solutions like rollups come in.

Off-chain scaling solutions

Rollups

Rollups are a solution that effectively scales Ethereum by compressing transaction data before it is posted to Layer 1.

Rollups execute a bundle of transactions off-chain before compressing the data and sending it back to Ethereum (on-chain), thus consuming less blockspace (data). One can think of a rollup as another scalable blockchain, but one that is tied to Ethereum via a smart contract.

As such, rollups inherit their security from Ethereum, while not being affected by the scalability trilemma. Rollups are able to sacrifice decentralization, while focussing on scalability.

However, the limited data capacity on Ethereum remains a bottleneck for rollups and will be increased through sharding on Layer 1, which we explore in detail later in this blog.

Different types of rollup

The transactions that take place on Layer 2 need to be proven as valid on Layer 1 using retroactive fraud proofs (optimistic rollups) or zero-knowledge proofs (zero-knowledge rollups), which compress data even more efficiently.

Optimistic rollups

Optimistic rollups move transaction execution off-chain, but post compressed transaction data to the Mainnet as calldata, thereby reducing fees for end-users.

The name “optimistic” derives from the assumption that these off-chain transactions are valid, so optimistic rollups do not publish their own proofs of validity for these transactions. This makes them different to zero-knowledge (ZK) rollups, which do publish cryptographic proofs of validity for off-chain transactions.

Optimistic rollups, instead, utilize retroactive fraud-proofs to detect cases where the transactions might not be executed correctly. Once a rollup batch is submitted on Ethereum, anyone can challenge the result of the transaction during a set time window (called a challenge period) by computing a fraud proof.

If this rollup batch remains unchallenged, after this challenge period, the transaction is considered as valid. Otherwise, the rollup transaction will be re-executed, with any incorrect transactions receiving a penalty.

Pros and cons of optimistic rollups

Those who want to avoid Ethereum’s high gas fees can opt for one of the existing Layer 2 solutions, such as Arbitrum and Optimism. Currently, transactions on those Layer 2 scaling solutions cost ten times less than the same transaction on Layer 1.

To use one of these options, users must deposit assets into the rollup’s bridge contract on Ethereum, which will forward the transaction to Layer 2, where an equal amount of tokens will be minted.

As optimistic rollups are designed to be compatible with the Ethereum Virtual Machine (EVM), developers can deploy their existing dApps to Layer 2 without great complications.

However, when withdrawing from an optimistic rollup to Ethereum, users have to allow around seven days for the challenge period to pass.

Zero-knowledge (ZK) rollups

Zk-rollups, such as StarkNet and Aztec Protocol, work similarly to optimistic rollups. However, unlike optimistic rollups, they publish cryptographic zero-knowledge proofs (validity proofs) validating that the transaction batches they have posted on Layer 1 are valid.

As such, they can process thousands of transactions in one batch, before they post a summary of all modifications needed to represent those transactions.

The validity proof is vital to prove that the transaction batch was executed correctly and to finalize them, eliminating the need to post all transaction data on-chain. Therefore, when transferring assets from Layer 2 to Layer 1, users are not subject to the same waiting period as with optimistic rollups.

As a hybrid solution that achieves huge scalability gains off-chain while inheriting on-chain security, ZK-rollups are considered much more secure than sidechains (Polygon PoS and Gnosis Chain (formerly xDai)) or validiums.

Validiums

Validiums are scaling solutions very similar to ZK rollups. They also use validity proofs to verify off-chain transactions on the main chain.

However, validium chains store state data off-chain. As a result, users could be deprived of their assets if this data is withheld, as it is required to prove ownership of funds and make withdrawals on Ethereum.

In addition, while validiums offer tremendous scalability, their approach to data storage has implications for their security and trust minimization.

Advantages of off-chain scaling: Lower transaction costs

A key advantage of moving execution and computation off-chain are lower fees. This is facilitated by the fact that rollups can specialize in a particular use case, unlike Ethereum itself, which needs to be able to execute all kinds of applications (decentralized exchanges, payments, NFTs, etc).

An example is Loopring — a ZK-rollup which is a decentralized exchange (DEX) tailored to maximize efficiency for its use case. As a result, the DEX is able to offer traders fees 30–100 times lower than on a DEX running on L1, such as Uniswap.

Limits of off-chain scaling

Rollups rely on Ethereum’s data capacity, aka blockspace. Although they use clever techniques to compress transaction data to reduce network congestion, this dependency means they can only offer linear scalability gains.

To further improve their off-chain scaling capacity, Ethereum needs to expand its data capacity. This leads us to on-chain scaling, which aims to remove this bottleneck.

Scaling Ethereum via on-chain scaling

Ethereum’s team of developers plans to increase Ethereum’s data capacity in two phases. The first to be introduced is proto-danksharding, which is already being prototyped, followed by danksharding itself.

What is danksharding?

First thing first, let’s look at what danksharding is. Without sharding, each transaction on Ethereum needs to be approved by all validators. With full dankharding, these validators would be used much more efficiently as they are distributed across data shards.

This is made possible by the PoS consensus mechanism introduced with the Merge, which allows PoS validators to be distributed across the data shards.

If 6,400 validator nodes were securing the Beacon chain before sharding, 100 validators would now secure each of the 64 data shards. Effectively, this would split the security budget (staked ETH) across multiple data chains.

Each shard is expected to provide 1/3 of the data capacity of Ethereum’s current Beacon chain. Consequently, 64 shards would increase Ethereum’s data capacity by 18 times, further multiplying scalability gains.

Without sharding or on-chain scaling, the scalability gains of rollups are linear, as they are ultimately limited by the amount of data they can deploy on Layer 1. As Ethereum removes this bottleneck through sharding, the efficiency gains of rollups will multiply.

It is believed that the maximum total throughput of rollups could increase from 5K transactions per second (TPS) to 100K TPS, based on a total of 64 data shards. However, in its final phase, Ethereum plans to split its data layer into up to 1,024 shards.

It is important to note that Ethereum’s roadmap is focused on rollups, as on-chain scaling solutions are specifically designed to scale Layer 2 solutions. While danksharding will provide scalability gains for Layer 1 transactions, these gains will multiply on Layer 2.

Consequently, to sustainably reduce network congestion, users will need to utilize Layer 2s for most of their transactions in the future, which will then be secured by Ethereum.

What is proto-danksharding (EIP-4844)?

Before danksharding, however, comes something called “proto-danksharding”. Proto-danksharding is a on-chain scalability solution specifically designed for rollups. It aims to further increase their scalability gains by providing them with more data capacity.

The name can be confusing, as proto-danksharding does not introduce sharding technology to Ethereum quite yet. Rather, it is a predecessor to full danksharding.

To increase data capacity for rollups, proto-danksharding introduces a new type of data called blob-data, providing much more and cheaper data to rollups compared to traditional calldata, which is currently used by rollups and applications on Ethereum.

As blobs are much larger and cheaper, this upgrade is expected to provide large scalability gains for Layer 2s. According to Optimism, the upgrade could result in up to 100x cheaper fees for transactions on optimistic rollups.

In addition, since the additional blob-data is reserved for rollups and is not accessible for transactions on Layer 1, rollups do not need to compete for blockspace with applications on Ethereum itself.

Adoption of scaling solutions

In order to contribute to a significant fee reduction on the main layer, a larger share of transactions will have to be moved to Layer 2. However, so far in 2022, only 0.5% to 1.5% of all fees incurred on Ethereum come from Layer 2. While this share remains small, it has increased significantly compared to just one year ago.

At this stage, many Layer 2 solutions are still nascent and do not yet offer as wide a range of applications as Ethereum itself. However, this is expected to change over time, as large projects such as Uniswap and Aave are already available on Optimism and Arbitrum.

Conclusion — Scaling the Ethereum blockchain

As the above outline shows, Ethereum’s journey is far from over. According to its founder Vitalik, Ethereum is only 55% done after the Merge.

Once this journey is complete, Ethereum is expected to handle 100,000 transactions per second. The good news is that rollup fees should drop below $0.05 by as early as 2023, once proto-danksharding is implemented.

So far, Ethereum has been fertile ground for one of the most innovative industries the world has ever seen. If the development team delivers on its promises, Ethereum could not only defend its status as the most secure smart contract platform, but also become the most scalable and accessible blockchain, while continuing to be the foundation for one of the most innovative technologies of our time. The team of Ethereum developers might be well on their way to creating the settlement layer of the emerging Web3 ecosystem and beyond.