Scaling Ethereum: past, present and future

In the first part of this two-part series, we looked at one of the most commonly asked questions about blockchain networks: how they can scale up to accommodate millions of transactions.

 

If we turn to Ethereum, we see a network that is under pressure from many different requirements, and which has been plagued with periods of congestion and high fees. Because the majority of DeFi protocols and NFT marketplaces run on Ethereum, it has in many ways become a victim of its own success.

 

Many people who would have otherwise wished to get started in DeFi have been deterred by high transaction fees, which have averaged around $20 for the last couple of months, rising to more than $100 at peak times. With more than 17 billion locked up in Ethereum DeFi protocols, the prospect of eye-wateringly high potential returns continues to attract large investors, but sky-high fees make it unprofitable for smaller players to participate. A collateralized loan of $1,000 makes little sense if you are paying transaction fees of up to $100 to open and then again to close the loan.

 

As well as high fees, unpredictable delays in processing transactions can be a problem. This is hardly a new phenomenon: at the height of the CryptoKitties craze in December 2017, one of the higher-profile casualties of Ethereum’s main net becoming clogged up with cartoon cats was SophiaTX, who were forced to delay their ICO. As well as high fees, unpredictable delays in processing transactions can be a problem. This is hardly a new phenomenon: at the height of the CryptoKitties craze in December 2017, one of the higher-profile casualties of Ethereum’s main net becoming clogged up with cartoon cats was SophiaTX, who were forced to delay their ICO. 

 

Congestion and high fees are unsurprising when we consider that Ethereum can currently process only around 15 transactions per second (500,000 or so per day) on its main network.

 

Fortunately, more than three years on, we are now in a situation where much work has been done to address these challenges. Because Ethereum is open-source code, which can be downloaded and run by anyone in the world, there are – and will continue to be – many different opinions about the best way to achieve scalability. In this respect, Ethereum is similar to Bitcoin. However, the most famous of Ethereum’s founders, Vitalik Buterin, is not anonymous like Satoshi Nakamoto, and this important difference means that his opinions and pronouncements heavily influence the opinions of developers, node validators and users.

 

Like Bitcoin, scaling Ethereum can be achieved in different ways: either by making changes to the underlying network itself, or by providing technical solutions that run on top of, or alongside, the main network. The latter are referred to as Layer 2 (or L2) solutions. While Ethereum’s roadmap encompasses an ambitious L1 redesign that will move it away from Proof of Work to Proof of Stake, Buterin has made it clear that he sees future work to increase the throughput of the network as dependent on both L1 and L2 developments.

 

In this interesting thread in Ethereum Magicians, he admits: “The L1 is nearly unusable for many classes of applications, and there’s no non-L2 path that can get us to scalability in the short-to-medium term”. In this, he acknowledges that the transition to Ethereum 2.0, whose timetable has repeatedly slipped, is far from being a silver bullet in the near future – and even in the best case scenario, will only reach its maximum potential throughput in combination with L2 innovationsIn this interesting thread in Ethereum Magicians, he admits: “The L1 is nearly unusable for many classes of applications, and there’s no non-L2 path that can get us to scalability in the short-to-medium term”. In this, he acknowledges that the transition to Ethereum 2.0, whose timetable has repeatedly slipped, is far from being a silver bullet in the near future – and even in the best case scenario, will only reach its maximum potential throughput in combination with L2 innovations

Ethereum’s move to Proof of Stake

Rather than being a Big Bang transition, the huge shift at the heart of Ethereum is happening in stages, with a whole new infrastructure being built alongside the original Ethereum network in a process that is taking, quite literally, years.

 

While Proof of Work is based on the principle that in order to secure the history of transactions on the network, each participating node must expend a certain amount of computational power to verify each block, Proof of Stake is based on a less energy-intensive model where the organisations or individuals who run the validating nodes propose and attest blocks. Consensys have a good explanation of the process here: “Validators… make themselves available to be randomly selected to propose a block. Other validators then “attest” that they have seen the block. When a sufficient number of attestations for the block has been collected, the block is added to the blockchain.”

 

Critics of Proof of Stake argue that while validators must stake a certain amount of crypto in order to participate in this process (their stake is locked in a smart contract and can be slashed for malicious or fraudulent behaviour), this amount of skin in the game cannot compete with Proof of Work in terms of security. Additionally, Proof of Stake brings the risk that the consensus becomes increasingly controlled by those who already hold the majority of assets. Its supporters contend that not only is a less energy-intensive validation process beneficial in itself, but also that the income generated by staking provides an incentive for more validators to join the network, thus reducing centralization.

 

Ethereum 2.0 promises more than simply a change to its consensus mechanism. A development called sharding (not a new idea – sharding is a technology that is already applied, for example, to databases) will allow for transactions to be validated on multiple separate shard chains, thus vastly increasing capacity. And over the entire five-to-ten year roadmap, other technologies are planned to be introduced that will mean that by 2030 Ethereum is barely recognisable.

 

The current state of play (March 2021) is that the Beacon chain (the backbone of Ethereum 2.0) was launched in December 2020, and now has more than three million ETH staked to secure it. The roadmap shown here indicates that sharding will arrive some time later this year, while the merging of the current Ethereum blockchain’s data into the new chain will probably happen in 2022. It is important to point out that neither is this a migration in the traditional sense, nor will the underlying ETH token change. Ethereum’s current network will simply become a shard of the new chain, and all existing transaction data will be available on Ethereum 2.0. It is also worth noting that smart contract execution capabilities will become available on the new network at a different time from simple data transfers. It is worth reading the material on ethereum.org and Consensys to get an overview of what is happening when.

Layer 2 solutions - Plasma chains

A sidechain is a secondary blockchain connected to the main blockchain via a two-way peg, which may have its own consensus protocol. Plasma chains are often referred to as sidechains, but in fact there are important differences, and Plasma chains are an architectural pattern unto themselves.

 

As the Plasma documentation itself explains, there are some key differences that make them more secure: “Just like sidechains, plasma chains have a consensus mechanism that creates blocks. However, unlike sidechains, the “root” of each plasma chain block is published to Ethereum. Block “roots” are basically little pieces of information that users can use to prove things about the contents of those blocks. For example, a user could use a block root to prove that they made a transaction in that specific block.”

 

Within Plasma, there are different architectural implementations, as can be seen by the differences between OMG and Matic Network, both of whom use Plasma. Matic (recently rebranded as Polygon, although Matic Network itself retains the same name) have a good explanation here of where Plasma can be useful and where it is less so: they use an adapted version of Plasma with proof of stake-based sidechains, which intermittently push checkpoints to the Ethereum main chain.

 

Matic’s speed has made it attractive to dApps across a wide spectrum of use cases, especially gaming, where it is used by the likes of Decentraland.

Layer 2 solutions - rollups

Rollups are an alternative, and over the last couple of years have attracted a great deal of interest and development resources. Rollups tackle the problem in a different way: by aggregating transactions inside smart contracts. They have the potential to increase throughput from 15 transactions per second to more than 1,000.

 

Vitalik Buterin has highlighted the importance of rollups in Ethereum’s future, regardless of the move to Ethereum 2.0. In a post on Ethereum Magicians, he emphasised the importance of continuing L2 innovation, even while the L1 roadmap is being completed: “ “If you have rollups, but you do not have sharding, you still have 100X factor scaling,” Buterin said. “You still have the ability for the blockchain to go up to somewhere between 1,000 and 4,000 transactions a second, depending on how complex these transactions are.”

 

In a further comment on the same post, he went so far as to say: “L1 is nearly unusable for many classes of applications, and there’s no non-L2 path that can get us to scalability in the short-to-medium term”, while also highlighting the fact that funding research into L2 solutions, for example via Gitcoin grants, can be less contentious from a public goods point of view, and therefore healthy for the sustainability of the Ethereum ecosystem as a whole https://ethereum-magicians.org/t/a-rollup-centric-ethereum-roadmap/4698.

 

There are two approaches to rollups:

  • Optimistic rollups, such as those developed by Optimism, base their security model on assuming transactions are valid by default, and run a fraud proof computation in the case of a challenge
  • Zero-knowledge rollups, for example, Loopring or Starkware, work by running their own computations off-chain and submitting validity proofs to the main chain.

If you want to read more on this, I’d recommend digging into the Layer 2 Scaling content at ethereum.org: https://ethereum.org/en/developers/docs/layer-2-scaling/

 

Either way, if Ethereum is going to thrive and compete with all the new competition that is emerging, Buterin is convinced that users need to get used to a completely new way of thinking about how they use the network, “Currently, users have accounts on L1, ENS names on L1, applications live entirely on L1, etc,” he said on Ethereum Magicians. “All of this is going to have to change. We would need to adapt to a world where users have their primary accounts, balances, assets, etc entirely inside an L2.”

 

The whole conversation around Ethereum and how its various interlocking layers will scale is not a straightforward one, and looks set to run for months and probably years. 

About Author
About Author