Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Chasing the Speed of Light: Low-Latency Planetary-Scale Adaptive Byzantine Consensus

Christian Berger, Lívio Rodrigues, Hans P. Reiser, Vinicius Cogo, Alysson Bessani

TL;DR

Mercury is presented, a novel transformation to autonomously optimize the latency of quorum-based BFT consensus and upholds standard SMR safety and liveness guarantees with optimal resilience, thanks to its judicious use of a dual operation mode and BFT forensics techniques.

Abstract

Blockchain technology sparked renewed interest in planetary-scale Byzantine fault-tolerant (BFT) state machine replication (SMR). While recent works predominantly focused on improving the scalability and throughput of these protocols, few of them addressed latency. We present Mercury, a novel transformation to autonomously optimize the latency of quorum-based BFT consensus. Mercury employs a dual resilience threshold that enables faster transaction ordering when the system contains few faulty replicas. Mercury allows forming compact quorums that substantially accelerate consensus using a smaller resilience threshold. Nevertheless, Mercury upholds standard SMR safety and liveness guarantees with optimal resilience, thanks to its judicious use of a dual operation mode and BFT forensics techniques. Our experiments spread tens of replicas across continents and reveal that Mercury can order transactions with finality in less than 0.4 seconds, half the time of a PBFT-like protocol (optimal in terms of number of communication steps and resilience) in the same network. Furthermore, Mercury matches the latency of running its base protocol on theoretically optimal internet links (transmitting at 67% of the speed of light).

Chasing the Speed of Light: Low-Latency Planetary-Scale Adaptive Byzantine Consensus

TL;DR

Mercury is presented, a novel transformation to autonomously optimize the latency of quorum-based BFT consensus and upholds standard SMR safety and liveness guarantees with optimal resilience, thanks to its judicious use of a dual operation mode and BFT forensics techniques.

Abstract

Blockchain technology sparked renewed interest in planetary-scale Byzantine fault-tolerant (BFT) state machine replication (SMR). While recent works predominantly focused on improving the scalability and throughput of these protocols, few of them addressed latency. We present Mercury, a novel transformation to autonomously optimize the latency of quorum-based BFT consensus. Mercury employs a dual resilience threshold that enables faster transaction ordering when the system contains few faulty replicas. Mercury allows forming compact quorums that substantially accelerate consensus using a smaller resilience threshold. Nevertheless, Mercury upholds standard SMR safety and liveness guarantees with optimal resilience, thanks to its judicious use of a dual operation mode and BFT forensics techniques. Our experiments spread tens of replicas across continents and reveal that Mercury can order transactions with finality in less than 0.4 seconds, half the time of a PBFT-like protocol (optimal in terms of number of communication steps and resilience) in the same network. Furthermore, Mercury matches the latency of running its base protocol on theoretically optimal internet links (transmitting at 67% of the speed of light).
Paper Structure (38 sections, 4 theorems, 6 equations, 13 figures)

This paper contains 38 sections, 4 theorems, 6 equations, 13 figures.

Table of Contents

  1. Introduction
  2. Smaller Quorums for Better Latency
  3. Challenges and the Big Picture
  4. Contributions
  5. Background
  6. Byzantine State Machine Replication
  7. Weighted Quorums in BFT Replication
  8. System Model and Design
  9. Mercury: Threshold-Adaptive BFT
  10. Challenge #1: Dealing with $f > t_{fast}$ Failures
  11. Safety
  12. Periodic Checkpoint
  13. Client Panic
  14. Liveness
  15. Finalization
  16. ...and 23 more sections

Key Result

Proposition 1

Let $o$ be an operation finalized in conservative mode in the $i$-th position of the decision log. After the system switches to subsequent fast mode, $o$ will still be the $i$-th operation executed in the system.

Figures (13)

  • Figure 1: Weighted quorums composition and resulting BFT SMR latency for different resilience thresholds ($t$) in our $n=21$ setup (see details in §\ref{['sec:evaluation']}).
  • Figure 2: Overview over BFT quorum systems for $n=21$ replicas.
  • Figure 3: Mercury two modes of operation ($t = 2$, $t_\mathit{fast} = 1$).
  • Figure 4: A summary of Mercury.
  • Figure 5: Lightweight forensics procedure.
  • ...and 8 more figures

Theorems & Definitions (4)

  • Proposition 1
  • Lemma 1
  • Theorem 1
  • Theorem 2