Leopard: Towards High Throughput-Preserving BFT for Large-scale Systems

With the emergence of large-scale decentralized applications, a scalable and efficient Byzantine Fault Tolerant (BFT) protocol of hundreds of replicas is desirable. Although the throughput of existing leader-based BFT protocols has reached a high level of 10
⁵
requests per second for a small scale of replicas, it drops significantly when the scale increases.This paper focuses on preserving high throughput as the BFT protocol’s scale is increasing. We identify and analyze a major bottleneck to leader-based BFT protocols due to the excessive workload of the leader at large scales. A new metric of scaling factor is defined to capture whether a BFT protocol will get stuck when the scale gets larger, which can be used to measure the performance of throughput and scalability of BFT protocols. We propose "Leopard", the first leader-based BFT protocol that scales to multiple hundreds of replicas, and more importantly, preserves high throughput. We remove the bottleneck by introducing a technique of achieving the ideal constant scaling factor, which takes full advantage of the idle resource and balances the workload of the leader among all replicas. We implemented Leopard and evaluated its performance compared to HotStuff, a state-of-the-art leader-based BFT protocol. We ran extensive experiments with up to 600 replicas. The results show that Leopard achieves significant throughput improvements. In particular, the throughput of Leopard remains at a high level of 10
⁵
when the scale is 600. It achieves a 5× throughput over HotStuff when the scale is 300, and the gap becomes wider as the scale further increases.