Background
Tristan (Host) – host at Y-Axis Podcast Keone Han (Guest) – CEO of Monad Monad – a fundamentally optimized EVM blockchain
Introduction and Background of Keone Han
Keone mentions his prior experience in high-frequency trading for about 10 years. He describes high-frequency trading as an industry that involves building high-performance systems and dealing with vast amounts of data. He talks about the scale of global financial systems and how exchanges would send hundreds of millions of messages daily. He highlights the need to build systems that could handle such data scales and respond in microseconds. He reflects on his experience at Jump Trading as valuable. He shares that he joined the crypto division of Jump Trading for a brief period. Keone mentions that this is how he and Tristan first connected. He talks about his work on Solana DeFi with his co-worker James , with whom he started working in 2014, and notes that both he and James were initially working on high-frequency trading together. He shares that after spending about seven or eight months in Jump crypto, they became very excited about decentralized computation. He says that this excitement led them to leave Jump Trading and start Monad together. He states that they have been working on Monad for approximately a year and a half.
Transition from High-Frequency Trading to Blockchain
Keone highlights the composability of decentralized apps, where one can build an app, deploy it, and then others can build upon it without permissionless. He emphasizes the atomic nature of blockchain functions, where a function in one app can be a subroutine in another app. He expresses excitement about the potential of blockchain technology, particularly the instant settlement feature, the self-custodial nature of assets, the ability to operate without trusting centralized entities, and the potential of decentralized applications to replace existing systems and eliminate middlemen, providing direct services accessible to anyone. He shares his motivation for working in the crypto space, revealing that after directly working on this at Jump Crypto, he and James felt compelled to continue in the crypto domain. He discusses the realization that their background in building high-performance systems was highly relevant to addressing challenges in the crypto space. They saw a need to make the Ethereum Virtual Machine more performant. He describes the transition as a confluence of interest, excitement, and the realization that their expertise could address pressing issues in the crypto world.
Introduction to Monad
Keone defines Monad as a high-performance, fully bytecode EVM-compatible layer one. He states that Monad introduces pipelining to the Ethereum Virtual Machine. He highlights that Monad aims to process transactions at a rate of 10,000 transactions per second. He points out that any application developed for Ethereum can be redeployed on Monad without any changes. Keone explains that the same bytecode from Ethereum can be redeployed on Monad without modifications. He describes Monad has multiple executors working in parallel, which enhances efficiency. He shares that in Monad, transactions can be processed in parallel, unlike the sequential processing in Ethereum. He notes that when interacting with a smart contract on Monad, the virtual machine pulls the bytecode and associated state from disk, then steps through individual opcodes to execute the function being called. This process is similar to Ethereum but is made more efficient in Monad due to parallel processing. He mentions that if two individuals were to deposit into a platform like Aave on Ethereum, the transactions would be ordered sequentially. In contrast, on Monad, these transactions would essentially be processed in parallel.
Pipelining in Monad
Keone introduces pipelining as a term from traditional computer science, commonly used in CPU execution .He describes pipelining in a CPU as the process of executing stages of work through multiple phases, allowing the CPU to execute multiple machine instructions in parallel. He expresses surprise that Ethereum and other EVM-compatible blockchains are essentially single-threaded, executing transactions one after the other. Keone shares that Monad introduces pipelining in four major areas to address this limitation: State access pipelining: Enhancing the accessibility of the state in terms of reading and writing from the disk. Parallel execution of individual transactions: Allowing multiple transactions within a block to be processed simultaneously. Pipelining between consensus and execution: Dividing the work into stages and running consensus and execution in parallel, albeit with a slight delay for the execution side. Pipelining within the consensus mechanism: Incorporating an element of parallel communication where multiple stages of the consensus protocol are communicated simultaneously. He highlights that combining these pipelining techniques allows Monad to process transactions efficiently, achieving a throughput of 10,000 transactions per second and substantially scaling the Ethereum Virtual Machine. Keone clarifies that pipelining is a foundational concept in traditional computer science, particularly in how CPUs execute instructions more efficiently. He notes that every modern CPU utilizes pipelining under the hood. He expresses that it seemed odd to the Monad team that Ethereum and similar blockchains were single-threaded, processing transactions sequentially. He shares that one of Monad’s architectural improvements is to allow the system to process transactions more efficiently. Keone mentions that as they worked on improving transaction processing, they identified other bottlenecks, such as accessing state variables of specific smart contracts. This led to further pipelining efforts, like asynchronous reads and writes for batched disk operations. He discusses another bottleneck where, during the block time, a small portion is allocated to execution because most of it is spent on consensus. This realization led to changes that moved the execution out, further enhancing efficiency.
Scaling with EVM community
Tristan questions about Monad’s reception by the EVM community and if there has been any pushback. Keone responds that the reception has generally been positive and supportive. He notes that when they discuss their work in person, people appreciate the uniqueness of their effort in rebuilding the EVM execution stack from the ground up. He highlights that Monad’s innovations are orthogonal to other projects, meaning they can coexist without conflict. He suggests that in the future, roll-ups could potentially be built using Monad’s execution stack, combining the best of both worlds. He emphasizes the importance of collaboration in the crypto space and how different teams can work together to address various challenges. Keone expresses appreciation for the open-source culture in the crypto developer community and the opportunity to collaborate with a larger group of contributors.
Comparing Ethereum and Monad
Tristan expresses interest in understanding the life cycle of an Ethereum transaction and how it compares to Monad. Keone begins by describing the life cycle of an Ethereum transaction: Users typically initiate transactions via websites. After clicking a button on a website, a transaction is prepared and presented to the user. The user signs the transaction using their wallet, which holds the private keys. Once signed, the transaction becomes a signed message and is sent to an RPC server of a node. The transaction enters the mempool , a pending state, and is propagated to other nodes via a peer-to-peer gossip protocol. The transaction remains in the mempool until it’s included in a block.
He says that Ethereum’s network consists of nodes that are perfect replicas of each other, maintaining the same state and smart contracts. He adds that These nodes stay in sync through a consensus mechanism, which allows them to agree on the contents of a block. Keone mentions that the consensus mechanism is robust, tolerating up to one-third of the nodes being malicious. He says that the purpose of this elaborate system is to maintain a shared global state across nodes, enabling decentralized apps to function without a centralized authority. Keone compares Ethereum to a virtual bank, where decentralized apps can emulate traditional banking functions. He says that Ethereum produces blocks every 12 seconds, and uses a proof-of-stake mechanism for civil resistance, determining how nodes choose leaders and vote. He adds that Ethereum has a gas limit of about a million gas per second, translating to roughly 10 transactions per second for typical transactions. Keone introduces Monad as a system similar to Ethereum in terms of block and transaction ordering. He highlights differences such as Monad produces blocks every second, it achieves finality within a second, compared to Ethereum’s 12 to 18 minutes, it is fully EVM bytecode compatible, meaning contracts from Ethereum can be deployed on Monad without changes, and it shares that transaction format and address space are the same as Ethereum, ensuring compatibility with tools like MetaMask .
Monad’s current positioning
Keone states they are in active R&D and have built most of the components. He expresses excitement about the results and the future, targeting a mainnet launch in mid-2024. He emphasizes the importance of thorough testing and ensuring the protocol’s safety and resilience against potential threats. He mentions potential collaboration opportunities with ambitious builders of applications due to Monad’s full EVM compatibility. Keone encourages interested individuals to check out their job listings and reach out. He acknowledges that there are other areas in the crypto space where they aren’t experts and are looking to integrate solutions from others.
Important Links
Watch the original episode Follow Keone Han on Twitter Follow Tristan Yver on Twitter Follow Monad on Twitter
Information
Medium: Youtube Show: The Y-Axis Podcast Show Title: Keone Hon: Scaling EVM, HFT Trading, Pipelining and Building a Layer 1 Ecosystem Show Date: September 18th, 2023