Fast and Interactive Byzantine Fault-tolerant Web Services via Session-Based Consensus Decoupling

TL;DR

This paper introduces a two-layer architecture for Byzantine fault-tolerant web services that separates interactive feedback from consensus finalization, achieving under 200ms responsiveness while maintaining security.

Contribution

The paper presents a novel session-based consensus decoupling architecture that significantly improves responsiveness in BFT web services without compromising security.

Findings

Layer 2 operations are four times faster than Layer 1.

System achieves under 200ms response time.

Effective in domains like metaverse requiring quick interaction and security.

Abstract

Byzantine fault-tolerant (BFT) web services provide critical integrity guarantees for distributed applications but face significant latency challenges that hinder interactive user experiences. We propose a novel two-layer architecture that addresses this fundamental tension between security and responsiveness in BFT systems. Our approach introduces a session-aware transaction buffer layer (Layer 2) that delivers immediate feedback to users through consensus simulation, while periodically committing batched operations to a fully Byzantine fault-tolerant consensus layer (Layer 1). By separating interactive operations from consensus finalization, our system achieves responsive user experiences of under 200ms, while maintaining strong BFT security guarantees. We demonstrate the efficacy of our architecture through a supply chain management implementation, where operators require both immediate feedback during multi-step workflows and tamper-proof record keeping. Our evaluation shows that our Layer 2 operations perform four times faster than the Layer 1 counterpart, while substantially preserving the end-to-end transaction integrity. Our approach enables BFT applications in domains previously considered impractical due to latency constraints, such as metaverse environments, where users require both responsive interaction and guaranteed state consistency.