- The paper introduces Conflux, a blockchain protocol using a DAG-based Nakamoto consensus to concurrently process blocks, significantly improving scalability over linear chains like Bitcoin.
- Conflux achieves a throughput of approximately 6400 transactions per second in tests, confirming transactions in 4.5 to 7.4 minutes, a vast improvement compared to Bitcoin's performance.
- The research identifies individual node processing capability, particularly bandwidth, as the primary bottleneck for scaling Conflux, highlighting implications for industries requiring high throughput.
Scaling Nakamoto Consensus to Thousands of Transactions per Second
The paper "Scaling Nakamoto Consensus to Thousands of Transactions per Second" presents a blockchain protocol called Conflux, designed to improve the scalability and efficiency of blockchain systems using a DAG-based consensus mechanism. The proposal addresses key limitations of traditional blockchain systems, such as Bitcoin, which rely on linear chains.
Overview of Conflux
Conflux represents an advancement in blockchain technology by allowing the concurrent processing of blocks, thereby increasing throughput and reducing confirmation times. The core idea involves organizing blocks into a Directed Acyclic Graph (DAG) rather than a linear chain, which facilitates the synchronization and integration of concurrent blocks. The relationship between blocks is maintained through parent edges and reference edges, and consensus is achieved by identifying a pivot chain that effectively partitions blocks into epochs. This DAG-based approach allows Conflux to achieve consensus on a block total order while simultaneously deriving a transaction total order.
Key Results and Performance
The paper outlines rigorous evaluations conducted on Amazon EC2 clusters with up to 20,000 full nodes. Conflux achieves transactional throughput of 5.76 GB per hour, translating to approximately 6400 transactions per second for typical Bitcoin transaction sizes. This is a substantial improvement over Bitcoin's typical 7 transactions per second, mainly because Conflux eliminates the consensus bottleneck by deferring transaction total ordering. Furthermore, transactions in Conflux are confirmed within approximately 4.5 to 7.4 minutes, significantly faster than other consensus mechanisms that rely on sequential processing.
Experimental Setup and Findings
The experiments reveal that while Conflux is no longer limited by consensus protocol throughput, the individual node processing capability, notably bandwidth, becomes the bottleneck. This suggests that further optimizations in node processing could yield even higher throughputs. When compared with Bitcoin and GHOST consensus approaches, Conflux exhibits superior performance, as these protocols discard concurrent blocks as forks resulting in lower block utilization ratios.
Implications and Future Directions
The implications of Conflux are extensive, particularly in areas requiring high transaction throughput such as financial services, supply chain management, and healthcare logistics. By enabling faster and more scalable blockchain operations, Conflux could lead to reduced transaction fees and improved user experiences. The paper outlines a clear path for further research into incentive mechanisms to economically encourage honest participation among nodes. Furthermore, exploration into boosting processing capability through hardware enhancements, as well as adapting the protocol for other platforms such as proof-of-stake systems, could unlock additional efficiencies.
Conclusion
Conflux demonstrates how DAG-based Nakamoto consensus can scale blockchain operations to thousands of transactions per second while ensuring transaction irreversibility with high confidence. This paper provides a blueprint for future blockchain protocols aiming to overcome current system bottlenecks, and it opens avenues for deploying blockchain solutions on a larger scale across various industries. While Conflux shows promising scalability, continued research and practical advancements are necessary to fully integrate such systems into global infrastructure efficiently.