Classification and Performance Comparative Analysis of Distributed Consensus Algorithms

Abstract

Distributed consensus algorithms are the core mechanism for ensuring data consistency and fault tolerance in distributed systems. Their evolution, from classical algorithms like Paxos and Raft to blockchain, reflects the diverse requirements for consistency, security, and performance in different application scenarios. This article analyzes the classification system of distributed consensus algorithms and divides them into two major categories based on fault models: Byzantine fault-tolerant and non-Byzantine fault-tolerant. On this basis, the system systematically organized the processes and characteristics of Paxos, Raft, Practical Byzantine Fault Tolerance (PBFT), and blockchain consensus algorithms, and performed a comparative analysis of two key performance indicators: throughput and consensus latency. Research has found that as the number of nodes increases, the performance of the traditional PBFT algorithm decreases significantly, while the Byzantine Fault Tolerant (TBFT) algorithm shows better scalability in both throughput and latency. Finally, this article combines the characteristics of various algorithms, summarizes their applicable scenarios, and provides theoretical basis and practical reference for the selection of consensus algorithms in distributed systems.