DRAFTEE: A self-adaptive framework for BFT-based consensus to meet fluctuating throughput demands under security constraint

Abstract

Byzantine Fault Tolerance (BFT) consensus protocols are known not to scale to a number of participants as large as mainstream public blockchain solutions. This is due to the way transactions are validated, which is performed by a leader that acts as a bottleneck and cause performance limitations. In BFT-based blockchain systems, a significant challenge involves effectively managing the trade-off between security and throughput performance. This trade-off suggests that while a higher number of validators enhances system security, it also diminishes throughput performance, and vice versa. Traditionally, blockchain systems fix the number of validators during the design phase, which lacks flexibility in adjusting the validator pool in real-time to optimize the aforementioned trade-off. This article presents DRAFTEE, a self-adaptive framework for BFT-based consensus designed to continuously regulate the validator pool to meet transaction throughput demands while upholding security requirements. DRAFTEE’s control mechanism involves determining the optimum number of validators at any given time and selecting the most suitable nodes to minimize network latency, while meeting the users’ (transaction) demand. DRAFTEE is implemented and experimentally evaluated (through simulation and Grid5000). The results demonstrate significantly improved performance, equivalent to double the previous maximum throughput demand with same number of validators. Moreover, the implementation allows for a remarkable increase, up to 250%, in the number of validators maintaining equivalent service quality. This optimization allows to increase the minimal number of validators, while preserving throughput demand.