Zaptos: Towards Optimal Blockchain Latency

Abstract: End-to-end blockchain latency has become a critical topic of interest in both academia and industry. However, while modern blockchain systems process transactions through multiple stages, most research has primarily focused on optimizing the latency of the Byzantine Fault Tolerance consensus component. In this work, we identify key sources of latency in blockchain systems and introduce Zaptos, a parallel pipelined architecture designed to minimize end-to-end latency while maintaining the high-throughput of pipelined blockchains. We implemented Zaptos and evaluated it against the pipelined architecture of the Aptos blockchain in a geo-distributed environment. Our evaluation demonstrates a 25\% latency reduction under low load and over 40\% reduction under high load. Notably, Zaptos achieves a throughput of 20,000 transactions per second with sub-second latency, surpassing previously reported blockchain throughput, with sub-second latency, by an order of magnitude.

• The paper presents a parallel pipelined architecture that cuts end-to-end blockchain latency by up to 40% and sustains 20,000 TPS.
• It employs optimistic execution and commit strategies to expedite transaction processing and maintain seamless data availability.
• The design integrates state certification into consensus, ensuring safety without added delay and setting a benchmark for next-gen blockchains.

An Overview of "Zaptos: Towards Optimal Blockchain Latency"

The paper "Zaptos: Towards Optimal Blockchain Latency" addresses a significant challenge in modern blockchain systems—optimizing end-to-end latency while preserving high throughput. This work is spearheaded by researchers affiliated with Aptos Labs, and it proposes a novel pipeline architecture named Zaptos, designed to tackle latency issues inherent in blockchain systems.

Background and Motivation

The increasing demand for scalable and efficient decentralized systems has intensified efforts to enhance blockchain performance, particularly in reducing transaction latency. Conventional approaches focus predominantly on improving the Byzantine Fault Tolerance (BFT) consensus mechanisms, yet these are not the sole contributors to latency. End-to-end latency, which encompasses the time from transaction submission to confirmation, includes additional stages such as client-validator communication, block execution, state certification, result storage, and communication back to the client. This paper acknowledges that these stages collectively influence latency, thus advocating for a more holistic approach to latency reduction.

Zaptos Architecture

Zaptos introduces a parallel pipelined architecture aimed at minimizing end-to-end latency. The design is grounded in the observation that most blockchain systems process information serially through multiple stages, limiting throughput and latency optimization. By restructuring these stages into a parallel approach, Zaptos strives to accelerate the processing pipeline.

Key Innovations:

1. Optimistic Execution: Zaptos initiates block execution immediately upon receiving block proposals, bypassing the need for prior consensus ordering. This speculative execution approach requires mechanisms to handle potential discrepancies if blocks are not eventually ordered.
2. Optimistic Commit: Upon concluding execution, Zaptos persistently commits the results into storage without awaiting state certification. This ensures that data availability is maintained even when state certification is pending.
3. Certification Optimization: To prevent unsafe state divergence, Zaptos piggybacks state certification onto the consensus protocol. By attaching state certification to the consensus process, Zaptos guarantees safety without additional latency overhead.

Results

In practical evaluations, Zaptos demonstrates a substantial reduction in latency, achieving a 25% reduction under low load and over 40% under high load when compared to the baseline Aptos blockchain architecture. Notably, Zaptos achieves a throughput of 20,000 transactions per second (TPS) with sub-second latency, setting a new benchmark in blockchain performance metrics.

Implications and Future Directions

The Zaptos architecture presents both theoretical and practical advancements for blockchain systems. Theoretically, it challenges the conventional serial processing approach and paves the way for redesigning pipelined architectures to exploit parallelism. Practically, it sets a precedent for future blockchain designs focused on achieving ultra-low latency without sacrificing throughput.

Future developments may explore adaptive techniques that dynamically adjust the level of optimistic execution and commit based on network conditions, further enhancing robustness against adversarial conditions. Moreover, as blockchain applications expand across industries requiring rapid transaction confirmations, such as decentralized finance (DeFi) and supply chain management, Zaptos could serve as a cornerstone for developing next-generation blockchain networks.

Conclusion

This paper represents a notable advancement in blockchain technology by expanding the horizon of optimization beyond consensus latency to include the complete transaction pipeline. Through its innovative parallel pipelining approach, Zaptos not only sets a new performance standard but also redefines the architectural possibilities in blockchain systems.