Simulating BFT Protocol Implementations at Scale

TL;DR

This paper proposes a simulation-based approach to evaluate the scalability and performance of BFT protocol implementations efficiently, accurately forecasting their behavior under various network conditions without costly real-world deployments.

Contribution

It introduces a resource-efficient simulation method that allows rapid and realistic evaluation of unmodified BFT protocols at scale, bypassing the need for extensive real-world testing.

Findings

Simulation accurately predicts BFT protocol performance.

Method enables scalable evaluation without code modifications.

Supports geographic and node count variations.

Abstract

The novel blockchain generation of Byzantine fault-tolerant (BFT) state machine replication (SMR) protocols focuses on scalability and performance to meet requirements of distributed ledger technology (DLT), e.g., decentralization and geographic dispersion. Validating scalability and performance of BFT protocol implementations requires careful evaluation. While experiments with real protocol deployments usually offer the best realism, they are costly and time-consuming. In this paper, we explore simulation of unmodified BFT protocol implementations as as a method for cheap and rapid protocol evaluation: We can accurately forecast the performance of a BFT protocol while experimentally scaling its environment, i.e., by varying the number of nodes or geographic dispersion. Our approach is resource-friendly and preserves application-realism, since existing BFT frameworks can be simply plugged into the simulation engine without requiring code modifications or re-implementation.