Monad: High Performance Culture

A Different Approach to Scaling EVM Blockchains

Existing Ethereum Virtual Machine (“EVM”) chains face several significant limitations that hinder their scalability and performance. At the core of these issues is the interdependence of consensus and transaction execution during block creation, a process that consumes substantial time and resources, resulting in slow and expensive block production. Despite these challenges, the blockchain industry has been reluctant to modify and optimize the virtual machine itself, leading to only minor improvements in throughput over time. This reluctance is partly due to the focus on Layer 2 scaling solutions, which has diverted researchers' attention from enhancing base layer performance to scaling chain deployments (Orbit chains, Superchain). Furthermore, the cost associated with scaling node hardware presents a significant obstacle to improving performance while maintaining a comparable level of decentralization. This issue is particularly pronounced in the Ethereum community which has steadfastly maintained its commitment to low hardware requirements for solo stakers, further complicating efforts to enhance the network's overall capacity and efficiency.

Monad represents the vision of hyperscaling EVM-based blockchain technology, driven by the experience of its founders, Keone Hon and James Hunsaker. With their background in designing high-frequency trading (HFT) systems at Jump Trading, they bring a wealth of experience in high performance computing to the blockchain space. Monad is positioned as a new Layer 1 blockchain that aims to address the limitations of existing EVM chains by offering a combination of high throughput and low latency. The platform boasts custom execution and consensus clients capable of processing 10,000 transactions per second with 1-second single slot finality. Despite these high performance characteristics, Monad aims to maintain relatively low hardware requirements and full EVM bytecode and RPC compatibility, making it accessible and familiar to the Ethereum community. If, as targeted, the platform can maintain sub-cent gas fees it will further enhance its appeal for users. These ambitious performance metrics are primarily attributed to several key innovations implemented across the consensus and execution layers: the MonadBFT consensus mechanism, deferred execution, parallel execution, and the specialized MonadDB backend. Together, these innovations form the foundation of Monad's approach to scaling the EVM while maintaining compatibility with the Ethereum ecosystem.

Consensus and MonadBFT

Monad's consensus mechanism, MonadBFT, is a custom BFT (Byzantine Fault Tolerant) algorithm derived from HotStuff consensus. MonadBFT introduces several key optimizations, including a reduction in communication rounds from three to two, as well as an optimistic design that allows consensus rounds to progress at actual network latency, thereby improving overall performance. In addition to MonadBFT, another critical design aspect of Monad's architecture is the separation of consensus and execution via “deferred” or "asynchronous" execution. This is achieved via pipelining, a method of parallelizing tasks by dividing them into smaller, concurrent steps. In Monad’s case, nodes first come to agreement on the ordering of transactions, and then begin executing those transactions in parallel while also initiating consensus on the next block. This is in contrast to Ethereum where consensus and execution are tightly coupled, thus requiring conservative time and gas limits to accommodate the interdependency between the two processes. As a result of this pipelining, the total time available for execution relative to block time is significantly increased on Monad.

Pipelining explained via laundry - top: naive; bottom: pipelined. Source: Prof. Lois Hawkes, FSU