Interpretation of the Sui Public Chain Core Distributed System Protocol White Paper

Mysten Labs recently updated the Sui Lutris White Paper, confirming the following key achievements after several months of testing:

1. Under the use of PTBs and 5K TPS, Sui can handle 140,000 to 150,000 operations per second, far exceeding the actual performance during peak conditions of the mainnet.

2. Even in the case where some validator nodes stop operating, Sui's final confirmation delay can still be maintained at below 0.5 seconds.

The White Paper details the operation mechanism of Sui, security proofs, and how external testers can reproduce relevant data in their own validations.

After the Sui mainnet went live, it has attracted the deployment of applications such as games and NFTs. The Sui Lutris technical report recently released by Mysten Labs introduces a distributed system that supports Sui, capable of maintaining low latency while ensuring high throughput and long-term stability.

Blockchain technology has developed for more than a decade since the birth of Bitcoin, with emerging applications such as games and NFTs continuously emerging. The industry is continually exploring methods to enhance blockchain efficiency, with a particular focus on high-load processing and real-time latency optimization.

Currently, L1 blockchains face two major challenges: achieving high throughput while maintaining low latency, and ensuring the long-term stability of the consensus protocol. These challenges can be addressed through the dynamic participation and configuration adjustment of validating nodes.

DAG-based consensus protocols (such as Narwhal/Bullshark used by Sui) are an effective way to increase throughput. These protocols support the simultaneous execution of a large number of transactions, making them suitable for application scenarios like gaming and NFTs. However, their few seconds of latency significantly impact common transaction operations.

On the other hand, non-consensus protocols show great potential in reducing latency and enhancing scalability. These protocols achieve fast transaction processing by eliminating the consensus step. However, they are only suitable for a limited category of simple operations, which restricts the expressiveness of smart contracts and makes it challenging to dynamically adjust the set of validation nodes.

Sui Lutris, as the core protocol of the Sui network, innovatively combines DAG-based consensus and non-consensus methods, achieving the advantages of both: sub-second latency and sustained throughput of thousands of transactions per second. At the same time, Sui retains the ability to execute complex contracts on shared objects, generate checkpoints, and reconfigure the set of validating nodes across cycles.

Sui Lutris adopts a unique hybrid approach. For operations on exclusive objects, the system utilizes a consistent broadcasting protocol among validation nodes to achieve latency below consensus. For complex smart contracts on shared objects, it relies on a consensus mechanism. Additionally, Sui Lutris supports network maintenance operations, such as defining checkpoints and reconfiguring validation nodes.

The transaction lifecycle of Sui Lutris includes the following steps:

1. The user creates and signs the transaction.
2. The validation node performs validity checks and signs.
3. The client collects the responses of most validating nodes to form the transaction certificate.
4. The certificate is sent back to the verification node for confirmation.
5. The consensus engine processes certificates and executes shared object operations.
6. The client collects the responses from the validation nodes as proof of transaction settlement.
7. Form checkpoints for each consensus submission.

Sui Lutris also provides a range of features supporting product-level blockchains, including checkpoint protocol, periodic reconfiguration, and mechanisms to resolve asset mislocking issues.

As the foundation of Sui, the complete technical report of Sui Lutris provides more details about the security and liveness protocol, as well as the security proofs in the standard distributed system model.