The Pipeline Knowledge Graph

Performance and scalability of blockchain systems

6 min read

From: thepipeline_xyz

Monad’s public testnet is launching as a showcase of the technology that has been built over several years, marking what is seen as a starting line rather than a finish line for innovation in blockchain performance [00:00:00], [00:00:26], [01:19:19], [01:29:47], [01:34:34], [01:48:47].

Monad’s Approach to Performance Optimization

Monad has taken a unique approach to blockchain development by building its protocol entirely from scratch [01:46:00]. This strategy provides more technical control over the product [02:06:00], allowing specific control over implementations like memory allocation and disk interaction to achieve the best performance [02:41:00]. While some open-source projects are used, Monad did not fork an existing client because it would have required too many modifications, negating the benefits of ongoing maintenance and new features [02:19:00], [03:11:00].

Previous efforts in EVM performance focused on areas like state sync performance rather than raw execution throughput [03:44:00], largely because Ethereum itself is arbitrarily rate-limited to a low throughput to maintain low node requirements [04:04:00]. Monad, however, was envisioned as a high-performance EVM chain with no design restrictions other than 100% EVM compatibility, aiming to maximize network and hardware performance [04:21:00], [04:31:00]. This vision also includes targeting consumer-grade nodes (e.g., a $1,000 PC) rather than expensive server machines [04:47:00].

Building a high-performance system from scratch is a massive undertaking, requiring specialized low-level system engineers who are difficult to find, as they typically work in big tech, silicon manufacturing, or high-frequency trading firms [05:12:00], [05:46:00].

Challenges in Blockchain Database Performance Optimization

Key challenges in Monad’s development involved achieving concurrent execution to leverage multi-core machines [06:17:00]. Unlike most existing clients that are single-core for transaction execution, Monad aims to keep multiple transactions in progress simultaneously [06:25:00], [06:34:00]. This requires intricate pipelining to manage dependencies between transactions within a block [06:42:00], [08:40:00].

A major innovation centers around the database. Disk access is a significant bottleneck, taking approximately 30 microseconds or more [07:10:00]. Sequential disk access quickly accumulates time, leading to idle CPU time [07:28:00]. Monad’s main innovations focus on executing transactions to always keep the machine busy, particularly by optimizing for disk access by building a custom database [07:53:00], [08:06:00]. This “pipelining” approach, likened to a washing machine and dryer analogy, allows for doing multiple steps across multiple objects at the same time, maximizing hardware utilization [08:13:00], [08:40:00].

Benchmarking High Throughput Blockchains

Benchmarks for Transactions Per Second (TPS) can often be misleading [00:07:00], [01:18:00]. Simple operations like native token transfers are the easiest for a blockchain to perform, and even unoptimized, single-core clients can achieve 50,000 TPS on such artificial benchmarks [01:51:00], [02:12:00]. ERC20 token transfers, while slightly more complex due to state updates, are still relatively simple [02:47:00].

Real-world Ethereum activity involves more taxing protocols like AMMs, lending, and especially expensive zero-knowledge (ZK) proofs, which have cost hundreds of dollars per transaction in the past [02:26:00], [02:34:00], [02:51:00]. Monad’s goal is to handle this “real usage” by replaying Ethereum history, which provides a realistic test of the system’s ability to cope with complex and expensive computations [01:51:00], [02:34:34], [02:50:00]. For example, one team on Monad’s devnet simulated 3 trillion gas usage, equivalent to about 30 days of Ethereum throughput, in just a few hours [01:56:00].

Misleading marketing in the blockchain space often involves manipulating benchmarking conditions, such as:

Monad rejects these practices, prioritizing technology and pushing the decentralization-performance tradeoff curve forward [03:25:00], [03:40:00]. They intentionally set up highly stake-weighted nodes in distant locations (e.g., Singapore and New York) to test and optimize the system under worst-case conditions, respecting the physical limitations of data transfer [03:43:00], [03:59:00]. Metrics that appear to violate the laws of physics are likely achieved through arrangements that compromise decentralization [04:35:00].

What High Performance Unlocks

Ultimately, a blockchain offers a shared global state where anyone can build and distribute applications globally [01:53:00], [03:06:00]. Monad aims to be a performant version of this, providing built-in payment rails and programmability to support millions or hundreds of millions of users with existing on-chain assets and digital identities [03:09:00], [03:15:00].

High performance enables several specific verticals:

  • High-Fidelity DeFi: Scaling personal finance to hundreds of millions of users with cheap transaction fees and low slippage, allowing liquidity providers to offer efficient on-chain markets [03:45:00], [04:07:00].
  • Consumer Space Applications: Any consumer-targeted application requiring scale to hundreds of millions of users and over a billion transactions per day, such as decentralized physical infrastructure networks (DePIN) [04:22:00], [04:26:00], [06:18:00].
    • This is especially relevant where applications may sponsor gas fees for users. Cheaper gas reduces the burden on developers, enabling sustainable business models [07:11:00]. Examples include marketplaces for health data, compute (CPU/GPU), cameras, and other device networks where consumers contribute resources [07:51:00].

The future of high-performance blockchains will see both predictable advancements, like more efficient DeFi, and unpredictable new applications, similar to how increased internet bandwidth led to YouTube and Instagram [08:39:00], [09:00:00]. Areas like AI agents and inscriptions are examples of emerging, unpredictable use cases that could thrive on high-throughput chains [09:26:00], [09:42:00].

The Evolution of Blockchain and Infrastructure Scaling

Monad’s devnet served as a feedback loop for early testers and infrastructure providers, helping to identify missing features or use patterns that required adjustments [07:35:00]. The public testnet is the next phase, providing a public product for everyone to use [07:46:00], [07:59:00].

The goal of the public testnet is to:

  • Exercise the technology and gather more feedback on the client’s strengths and weaknesses [08:45:00].
  • Showcase the power of the technology and the applications built on top of Monad that leverage its performance [08:37:00].
  • Allow the community to use and test the chain in real-world scenarios, including high-stress events like NFT mints or concurrent transactions from geographically dispersed users, to see how the system handles nearing its limits gracefully [09:16:00], [09:28:00]. A smooth degradation experience, where transactions might be slightly delayed but still go through, is preferred over obscure errors [09:41:00], [10:07:00].

Monad recognizes that launching the network is just the beginning. The core team views it as “crossing the starting line” rather than the finish line [00:26:00], [01:48:47]. There is a “massive queue of ideas” for future optimizations, rewrites to improve existing components, and ongoing research [04:59:00]. This includes new features for usability, privacy, and further decentralization, with a plan to scale the client to support thousands of nodes, on par with Ethereum and Solana [04:22:00], [04:31:00], [04:53:00], [04:59:00]. The long-term vision involves continuous innovation, moving beyond just delivering high performance and EVM compatibility to introduce many additional improvements [05:11:00], [05:23:00], [05:29:00].

Graph View

Backlinks