Blockchain performance optimization

From: thepipeline_xyz

High performance blockchains are a key focus in the development of new protocols, aiming to maximize throughput and efficiency while maintaining decentralization and compatibility. Monad’s approach involves building from scratch to achieve unparalleled control and optimization strategies for blockchain clients [00:02:04].

Building from Scratch for Performance

Monad chose to build its protocol entirely from scratch rather than forking an existing open-source project [00:01:46]. This approach provides greater technical control over the product [00:02:04].

Key reasons for this decision include:

• Technical Control: Building from scratch allows developers to be very particular about how components like memory allocation and disk interaction are implemented, ensuring optimal performance optimizations [00:02:41].
• Avoidance of Friction: If Monad had forked an existing project, such extensive modifications would have been needed that leveraging ongoing new features or maintenance from the original project would not have been possible [00:02:19].
• Long-term Maintenance & Feature Development: This approach offers simplicity, control, and better handling of ongoing maintenance responsibilities and new feature development [00:03:00].

While some open-source projects are used, Monad specifically avoided forking an existing client [00:03:11].

Challenges in Achieving High Performance

Achieving high performance blockchains presents significant challenges, particularly when dealing with concurrent execution and data access [00:06:09].

• Concurrency: Modern machines have many cores, but most existing blockchain clients are single-core for transaction execution [00:06:22].
  • Achieving parallel execution in blockchain is difficult due to dependencies between transactions within a block [00:06:36]. A block with 1,000 transactions is not 1,000 independent tasks [00:06:42].
  • Proper pipelining, where multiple transactions are worked on simultaneously, is crucial [00:06:47].
• Database Performance: The database is one of the most critical components for performance [00:07:10].
  • Waiting for disk access, which takes 30 microseconds or more, can quickly accumulate time if operations are sequential [00:07:12].
  • This sequential waiting prevents the full utilization of hardware resources, as the CPU sits idle while waiting for disk I/O [00:07:45]. This highlights challenges of standard databases in blockchain performance [00:07:10].

Monad’s Innovations in Performance

Monad’s main innovations focus on continuously keeping the machine busy and preventing the CPU from waiting [00:07:53].

• Custom Database: Monad built its own database (Monad DB) specifically to address the bottlenecks caused by disk access [00:08:10]. This is a key unique database optimization in blockchains [00:08:10].
• Pipelined Architecture: Monad employs a highly pipelined architecture, akin to a “washing machine” analogy [00:08:15]. Just as one might wash, dry, and fold different loads of laundry concurrently, Monad processes multiple steps of different transactions simultaneously [00:08:26]. This contrasts with simple parallel execution where transactions run side-by-side, instead opting for an offset, pipelined approach [00:08:50].
• Resource Utilization: The goal is to always keep the CPU doing something, optimizing as much as possible to maintain busyness [00:08:02].

“The difference between building a system that works and a system that’s really optimized… it’s like going from one to one hundred.” [00:09:23]

Benchmarking and Real-World Usage

Monad emphasizes benchmarking with real-world transaction data, specifically replaying Ethereum Virtual Machine optimizations (EVM) history, to ensure performance under actual user conditions [00:19:31].

• Misleading Metrics: Many projects report high Transactions Per Second (TPS) figures, but these often involve simple native token transfers or ERC-20 transfers [00:20:01].
  • Native transfers are the simplest for a blockchain to do, with some single-core clients achieving 50,000 TPS on such artificial benchmarks [00:19:51].
  • More complex transactions, like those involving AMMs, lending protocols, or expensive ZK proofs, are far more taxing on the system [00:21:26].
• Monad’s Standard: Monad aims to handle real usage, even optimizing for historically expensive computations like ZK proofs [00:22:34].
• False Marketing: The space often suffers from misleading marketing regarding performance metrics [00:22:41]. Tricks include:
  • Manipulating the nature of transfers or transactions [00:22:56].
  • Using high-end hardware for benchmarking [00:23:02].
  • Centralizing stake weight in one geographic location to speed up consensus, despite having nodes distributed globally [00:23:04].
  • Violating the laws of physics, implying unrealistic data transmission times [00:26:35].
• Monad’s Approach to Robustness: Monad intentionally sets up highly stake-weighted nodes in diverse geographical locations (e.g., Singapore, New York) to test and optimize performance under worst-case conditions [00:25:40].

Unlocking New Applications

Blockchain scalability and high-performance systems unlock new possibilities for developers and users. Monad envisions itself as a high-performance blockchain that offers shared global state with built-in payment rails and programmability [00:13:06].

This infrastructure is designed to support millions or hundreds of millions of users who already have assets and digital identities on-chain [00:13:20].

Specific verticals enabled by Monad’s performance include:

• High-Fidelity DeFi: Enabling personal finance at scale with cheap transaction fees and low slippage, leading to more efficient markets [00:13:45].
• Consumer Space: Supporting applications that need to scale to hundreds of millions of users with over a billion transactions per day [00:14:22]. This includes areas like DePIN (Decentralized Physical Infrastructure Networks) with price-sensitive data pushing [00:38:18].
• Simplified User Experience: Monad’s low transaction costs make it easier for applications to sponsor gas fees, abstracting away a significant barrier for users and enabling sustainable business models for developers [00:37:28].

Public Testnet: The Starting Line

The launch of Monad’s public testnet is seen not as a finish line, but as the “crossing the starting line” for continued innovation and growth [00:00:26].

• Devnet vs. Testnet: The devnet served as an early feedback loop for infrastructure providers to identify missing features and usage patterns [00:17:22]. The public testnet is a new phase, allowing broader community participation and testing [00:17:47].
  • During devnet, a team sent 3 trillion gas of usage (equivalent to about 30 days of Ethereum throughput) in a few hours, simulating complex exchange orders [00:17:53].
• Anticipated Use Cases: The testnet will reveal unanticipated use patterns [00:28:03]. Examples include:
  • Large-scale NFT mints [00:27:50].
  • High concurrency from geographically spread users [00:27:45].
  • Potential for AI agents to interact with the network [00:30:40].
• Graceful Degradation: The goal is for the system to handle stress gracefully and degrade smoothly rather than failing outright or providing poor user experiences (e.g., reverted transactions, RPC errors) [00:29:07].
• Long-Term Vision: Monad has a “massive queue of ideas” for future optimizations, rewrites for even better performance, new features for usability and privacy, and increased decentralization to support thousands of nodes [00:48:59]. This journey is seen as a long-term evolution, akin to constantly reinventing a high-performance trading system [00:50:47].
• Beyond Go-to-Market: Unlike many projects that focus on marketing existing codebases, Monad believes in bringing real, innovative technology to market as the start of a broader effort to drive crypto adoption [00:52:50].