Ethereum has been focused on scalability with The Merge as a major switch. Co-founder Vitalik Buterin has once again touched upon the topic, underlining that the roadmap is now rollup-centric.
This means that Ethereum is not centered on just sharding and layer 2 protocols have a crucial role for the main layer. Achieving trustlessness and interoperability is another essential task for this goal. All these have to be done while maintaining scalability, network security, and decentralization.
Ethereum is focusing on a rollup-centric roadmap
Ethereum co-founder Vitalik Buterin revisited the 2023 roadmap to discuss the network’s future. He said that Ethereum’s plan to scale has evolved from sharding and layer 2 protocols to a rollup-centric roadmap, which is now its current focus. Buterin explained that rollups will divide tasks between the Ethereum main layer and L2s.
Due to upgrades like EIP-4844, Ethereum L1 is now handling more data and different rollups (L2s) are working like shards, the co-founder explained.
He also says that Ethereum will finish developing the rollup-based scaling solution and work around challenges to the network’s security and decentralization intact.
Meanwhile, The Surge, which refers to a series of upgrades to improve Ethereum’s scalability and efficiency, has been assigned 4 main goals.
The first aim is to bring the transactions per second (TPS) at 100,000+ on layer 1 and layer 2 combined. The second is to preserve the decentralization and robustness of the main network.
The third is to ensure some L2s fully inherit Ethereum’s core features: being trustless, open, and resistant to censorship. Finally, it wants to promote maximum interoperability between L2s, so Ethereum feels like one ecosystem, not fragmented into separate blockchains.
To achieve these goals, Buterin clarifies that breaking the scalability trilemma is hard but not impossible. Buterin puts out 2 approaches to solve the trilemma of decentralization, scalability, and security.
The first approach is data availability sampling with SNARKs which will allow nodes to check that data is available and computations are correct. SNARKs (Succinct Non-Interactive Arguments of Knowledge) are trustless and remove the need for verification. The next approach is Plasma architecture, which involves offloading transaction processing and data checks to users in a smart, incentive-compatible way.
Implementation of PeerDAS comes with trade-offs
Buterin suggests that solving the trilemma will need advanced methods like SNARKs and Plasma and not just simple software fixes. With that, PeerDAS and other improvements also aim to make Ethereum more scalable by distributing data across the network.
While the Dencun upgrade improved data availability by introducing 375 kB of data per 12-second slot, Buterin says that it still limits the number of transactions.
In the case of Peer Data Availability Sampling, it uses “blobs” of data points. An Ethereum node reconstructs the entire blob using only a subset of these samples. Then in 1D Sampling, each node listens to a specific part of the data being broadcasted. Nodes can share the data samples without having to download all the data.
Yo @VitalikButerin, Very informative article. few questions for you though.
If rollups and EIP-4844 are the way to scale ETH, how’s DAS gonna keep it trustless without messing up decentralization? and what happens if all these L2s turn into their own little silos? How’s ETH gonna…— Zeerabets (@Zeerabets) October 17, 2024
The Ethereum co-founder explained that while PeerDAS allows nodes to request samples from others outside their subnet, there is a more conservative approach to using SubnetDAS using the local subnet.
With that, the blog underlines that the goal is to increase the data per slot to 16 MB, which would result in a potential throughput of around 58,000 TPS. Buterin writes, “And so ultimately we want to go further, and do 2D sampling, which works by random sampling not just within blobs, but also between blobs.”
Buterin underlines that the next step is rolling out PeerDAS and then increasing the blob count. However, the long-term goal is to reportedly move beyond 1D DAS and develop 2D Data Availability Sampling (DAS) to allow sampling both within and between blobs.
Buterin is also of the view that Ethereum needs to move away from the mathematical technique of KZG commitments and look toward a quantum-resistant, trusted-setup-free alternative like more advanced STARKs.
In the meantime, Ethereum needs to choose between 2D DAS and 1D DAS. There is a trade-off in both cases between efficiency and simplicity. The former offers the highest scalability but is more complex, while the latter is easier to manage but has a lower data capacity.
However, Buterin points out that Plasma could reduce the need for DAS. In this case, the trade-off would be a change in Ethereum’s architecture as reliance on users to handle data would increase. Buterin also suggests that if data compression techniques are implemented, it could reduce the need for advanced 2D DAS.
Ethereum can look at data compression technique
Data compression looks at cutting the size of each transaction on-chain to improve scalability. Ethereum could use different methods of compression like zero-byte compression, signature aggregation,
replacing addresses with pointers and custom serialization for transaction values.
Irrespective of the method used, the technique wants to make each transaction smaller in terms of bytes to process many more transactions within the same data space.
This will be a benefit but there are trade-offs in this case as well. For instance, Buterin explains that data compression would need switching to BLS Signatures, which will need major development. Replacing addresses with pointers is also complex for the client code. Overall, these changes could potentially make it harder to verify or audit transactions and could create problems for tools that rely on full transaction data.
That said, Buterin explains that all the current scalability solutions struggle in various aspects. Therefore, the Plasma scaling solution could become powerful as it utilizes an operator to publish blocks off-chain while putting the Merkle roots of those blocks on-chain.
Of course, bringing them to production remains a huge task. Achieving dependency on L2 proof systems is another problem statement for Ethereum. Buterin says that most rollups lack full trustlessness due to the presence of a security council.
With the main aim being achieving Stage 2 status, Buterin suggests formal verification techniques along with creating multiple proof systems and employing a multisig mechanism. Since advanced L2 proof systems will help reduce MEV stress on layer 1, scalability will be achieved as a result.
Among these developments, improvements in cross-L2 interoperability remain important. Buterin says that better interaction between L2s and L1 will enhance Ethereum’s overall usability. Unlike the other suggested changes, interoperability is not one technical upgrade. However, it is crucial to Ethereum’s long-term scalability roadmap.
