Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

JUMBO: Fully Asynchronous BFT Consensus Made Truly Scalable

Hao Cheng, Yuan Lu, Zhenliang Lu, Qiang Tang, Yuxuan Zhang, Zhenfeng Zhang

TL;DR

A signature-free asynchronous BFT consensus FIN-NG is proposed that adapts a recent signature-free asynchronous common subset protocol FIN into the state-of-the-art framework of concurrent broadcast and agreement, and a ``fairness'' patch is proposed for JUMBO, thus preventing a flooding adversary from controlling an overwhelming portion of transactions in its output.

Abstract

Recent progresses in asynchronous Byzantine fault-tolerant (BFT) consensus, e.g. Dumbo-NG (CCS' 22) and Tusk (EuroSys' 22), show promising performance through decoupling transaction dissemination and block agreement. However, when executed with a larger number $n$ of nodes, like several hundreds, they would suffer from significant degradation in performance. Their dominating scalability bottleneck is the huge authenticator complexity: each node has to multicast $\bigO(n)$ quorum certificates (QCs) and subsequently verify them for each block. This paper systematically investigates and resolves the above scalability issue. We first propose a signature-free asynchronous BFT consensus FIN-NG that adapts a recent signature-free asynchronous common subset protocol FIN (CCS' 23) into the state-of-the-art framework of concurrent broadcast and agreement. The liveness of FIN-NG relies on our non-trivial redesign of FIN's multi-valued validated Byzantine agreement towards achieving optimal quality. FIN-NG greatly improves the performance of FIN and already outperforms Dumbo-NG in most deployment settings. To further overcome the scalability limit of FIN-NG due to $\bigO(n^3)$ messages, we propose JUMBO, a scalable instantiation of Dumbo-NG, with only $\bigO(n^2)$ complexities for both authenticators and messages. We use various aggregation and dispersal techniques for QCs to significantly reduce the authenticator complexity of original Dumbo-NG implementations by up to $\bigO(n^2)$ orders. We also propose a ``fairness'' patch for JUMBO, thus preventing a flooding adversary from controlling an overwhelming portion of transactions in its output.

JUMBO: Fully Asynchronous BFT Consensus Made Truly Scalable

TL;DR

A signature-free asynchronous BFT consensus FIN-NG is proposed that adapts a recent signature-free asynchronous common subset protocol FIN into the state-of-the-art framework of concurrent broadcast and agreement, and a ``fairness'' patch is proposed for JUMBO, thus preventing a flooding adversary from controlling an overwhelming portion of transactions in its output.

Abstract

Recent progresses in asynchronous Byzantine fault-tolerant (BFT) consensus, e.g. Dumbo-NG (CCS' 22) and Tusk (EuroSys' 22), show promising performance through decoupling transaction dissemination and block agreement. However, when executed with a larger number of nodes, like several hundreds, they would suffer from significant degradation in performance. Their dominating scalability bottleneck is the huge authenticator complexity: each node has to multicast quorum certificates (QCs) and subsequently verify them for each block. This paper systematically investigates and resolves the above scalability issue. We first propose a signature-free asynchronous BFT consensus FIN-NG that adapts a recent signature-free asynchronous common subset protocol FIN (CCS' 23) into the state-of-the-art framework of concurrent broadcast and agreement. The liveness of FIN-NG relies on our non-trivial redesign of FIN's multi-valued validated Byzantine agreement towards achieving optimal quality. FIN-NG greatly improves the performance of FIN and already outperforms Dumbo-NG in most deployment settings. To further overcome the scalability limit of FIN-NG due to messages, we propose JUMBO, a scalable instantiation of Dumbo-NG, with only complexities for both authenticators and messages. We use various aggregation and dispersal techniques for QCs to significantly reduce the authenticator complexity of original Dumbo-NG implementations by up to orders. We also propose a ``fairness'' patch for JUMBO, thus preventing a flooding adversary from controlling an overwhelming portion of transactions in its output.
Paper Structure (28 sections, 16 theorems, 23 figures, 2 tables)

This paper contains 28 sections, 16 theorems, 23 figures, 2 tables.

Table of Contents

  1. Introduction
  2. Scalability Obstacles of the State of the Art
  3. Our Contribution
  4. Challenges and Our Techniques
  5. Challenges and Techniques of $\mathsf{FIN}\textrm{-}\mathsf{NG}$
  6. Challenges and Techniques of $\mathsf{JUMBO}$
  7. Problem Formulation and Preliminaries
  8. System and Threat Models
  9. Design Goal: Asynchronous BFT Consensus with Fairness
  10. Complexity Measurements
  11. Preliminaries
  12. Initial attempt: $\mathsf{FIN}\textrm{-}\mathsf{NG}$ to scale up Signature-free Asynchronous BFT
  13. $\mathsf{FIN}\textrm{-}\mathsf{NG}$ Protocol
  14. FIN-$\mathsf{MVBA}$-Q: $\mathsf{MVBA}$ with Optimal Quality
  15. Whether is $\mathsf{FIN}\textrm{-}\mathsf{NG}$ sufficiently scalable or NOT?
  16. ...and 13 more sections

Key Result

Theorem 1

$\mathsf{FIN}\textrm{-}\mathsf{NG}$ securely realizes asynchronous BFT atomic broadcast without fairness except with negligible probability, conditioned on the underlying w$\mathsf{BRBC}$ and $\mathsf{MVBA}$ protocols are secure (where $\mathsf{MVBA}$ shall satisfying quality).

Figures (23)

  • Figure 1: $\mathsf{Tusk}$tusk and $\mathsf{Dumbo}\textrm{-}\mathsf{NG}$gao2022dumbo for 16-256 nodes in LAN. Each node is an EC2 c6a.2xlarge instance with 12.5 Gbps bandwidth (upload+download).
  • Figure 2: Cost breakdown of Dumbo-NG in LAN consisting of 16-196 nodes.
  • Figure 3: CPU time breakdown of FIN for 16-196 nodes in LAN (when system load is close to peak throughput). Here "MT" represents Merkle tree merkle1987digital; "RS" represents Reed-Solomon code reed1960polynomial (instantiated by a fast implementation rs-lib of a systematic Cauchy Reed-Solomon variant blomer1995xorplank2005optimizing).
  • Figure 5: The paradigm of parallel broadcasts and agreement (the high-level rationale behind both $\mathsf{FIN}\textrm{-}\mathsf{NG}$ and $\mathsf{JUMBO}$).
  • Figure 6: High-level of $\mathsf{FIN}\textrm{-}\mathsf{NG}$.
  • ...and 18 more figures

Theorems & Definitions (34)

  • Definition 1
  • Definition 2
  • Definition 3
  • Theorem 1
  • Definition 4
  • Theorem 2
  • Definition 5
  • Theorem 3
  • Lemma 1
  • proof
  • ...and 24 more