Table of Contents
Fetching ...

Mysticeti: Reaching the Limits of Latency with Uncertified DAGs

Kushal Babel, Andrey Chursin, George Danezis, Anastasios Kichidis, Lefteris Kokoris-Kogias, Arun Koshy, Alberto Sonnino, Mingwei Tian

TL;DR

The paper introduces Mysticeti-C, a DAG-based Byzantine consensus that achieves the $3$-round latency lower bound by using uncertified blocks and a novel commit rule, and Mysticeti-FPC, a fast-path extension that embeds fast-path votes in the DAG to minimize signatures and boost throughput. It proves safety and liveness under Byzantine faults and evaluates both protocols against state-of-the-art baselines, showing sub-second latency and high throughput, including production deployment in Sui with over $1.5$B in value and notable latency reductions. The key contributions are the uncertified-DAG design with skip/certificate patterns, the multi-slot proposer model with a direct/indirect decision rule, and the integrated fast-path finality and epoch-change mechanism that preserve safety across reconfigurations. The practical impact is substantial latency reduction and resource efficiency for real-world blockchains, enabling high-throughput, low-latency consensus in geo-distributed deployments and production-ready deployments like Sui.

Abstract

We introduce Mysticeti-C, the first DAG-based Byzantine consensus protocol to achieve the lower bounds of latency of 3 message rounds. Since Mysticeti-C is built over DAGs it also achieves high resource efficiency and censorship resistance. Mysticeti-C achieves this latency improvement by avoiding explicit certification of the DAG blocks and by proposing a novel commit rule such that every block can be committed without delays, resulting in optimal latency in the steady state and under crash failures. We further extend Mysticeti-C to Mysticeti-FPC, which incorporates a fast commit path that achieves even lower latency for transferring assets. Unlike prior fast commit path protocols, Mysticeti-FPC minimizes the number of signatures and messages by weaving the fast path transactions into the DAG. This frees up resources, which subsequently result in better performance. We prove the safety and liveness in a Byzantine context. We evaluate both Mysticeti protocols and compare them with state-of-the-art consensus and fast path protocols to demonstrate their low latency and resource efficiency, as well as their more graceful degradation under crash failures. Mysticeti-C is the first Byzantine consensus protocol to achieve WAN latency of 0.5s for consensus commit while simultaneously maintaining state-of-the-art throughput of over 200k TPS. Finally, we report on integrating Mysticeti-C as the consensus protocol into the Sui blockchain, resulting in over 4x latency reduction.

Mysticeti: Reaching the Limits of Latency with Uncertified DAGs

TL;DR

The paper introduces Mysticeti-C, a DAG-based Byzantine consensus that achieves the -round latency lower bound by using uncertified blocks and a novel commit rule, and Mysticeti-FPC, a fast-path extension that embeds fast-path votes in the DAG to minimize signatures and boost throughput. It proves safety and liveness under Byzantine faults and evaluates both protocols against state-of-the-art baselines, showing sub-second latency and high throughput, including production deployment in Sui with over B in value and notable latency reductions. The key contributions are the uncertified-DAG design with skip/certificate patterns, the multi-slot proposer model with a direct/indirect decision rule, and the integrated fast-path finality and epoch-change mechanism that preserve safety across reconfigurations. The practical impact is substantial latency reduction and resource efficiency for real-world blockchains, enabling high-throughput, low-latency consensus in geo-distributed deployments and production-ready deployments like Sui.

Abstract

We introduce Mysticeti-C, the first DAG-based Byzantine consensus protocol to achieve the lower bounds of latency of 3 message rounds. Since Mysticeti-C is built over DAGs it also achieves high resource efficiency and censorship resistance. Mysticeti-C achieves this latency improvement by avoiding explicit certification of the DAG blocks and by proposing a novel commit rule such that every block can be committed without delays, resulting in optimal latency in the steady state and under crash failures. We further extend Mysticeti-C to Mysticeti-FPC, which incorporates a fast commit path that achieves even lower latency for transferring assets. Unlike prior fast commit path protocols, Mysticeti-FPC minimizes the number of signatures and messages by weaving the fast path transactions into the DAG. This frees up resources, which subsequently result in better performance. We prove the safety and liveness in a Byzantine context. We evaluate both Mysticeti protocols and compare them with state-of-the-art consensus and fast path protocols to demonstrate their low latency and resource efficiency, as well as their more graceful degradation under crash failures. Mysticeti-C is the first Byzantine consensus protocol to achieve WAN latency of 0.5s for consensus commit while simultaneously maintaining state-of-the-art throughput of over 200k TPS. Finally, we report on integrating Mysticeti-C as the consensus protocol into the Sui blockchain, resulting in over 4x latency reduction.
Paper Structure (29 sections, 21 theorems, 11 figures, 1 table, 3 algorithms)

This paper contains 29 sections, 21 theorems, 11 figures, 1 table, 3 algorithms.

Key Result

Lemma 1

If at a round $x$, $2f+1$ blocks from distinct authorities certify a block $B$, then all blocks at future rounds ($>x$) will link to a certificate for $B$ from round $x$.

Figures (11)

  • Figure 1: P50 latency of the Sui blockchain sui switching from Bullshark (1.9s) to Mysticeti-C (400ms) on 106 independently run validators
  • Figure 2: Block $(A_3, r+2, \cdot)$ (green) may reference blocks from different validators that support both $(A_3, r, L_r)$ (blue) and $(A_3, r, L_r')$ (red) equivocating blocks. If any of the blocks gathers $2f+1$ support, it will be certified, and we show that at most one may do so.
  • Figure 3: Illustration of main DAG patterns identified by validators.
  • Figure 4: Example application of the Mysticeti-C decision rule with four validators (A0, A1, A2, A3) and four proposer slots per round.
  • Figure 5: Illustration fast path transaction execution. The blocks $(A_0, r, \cdot)$ contain the fast path transactions $T_1$, $T_3$, and $T_6$. Blocks $(A_0, r+1, \cdot), (A_1, r+1, \cdot), (A_2, r+1, \cdot)$ support $(A_0, r, L_r)$ and explicitly vote for $T_1$ and $T_3$ (but not $T_6$). Upon observing these blocks, the validator can safely execute $T_1$ and $T_3$.
  • ...and 6 more figures

Theorems & Definitions (40)

  • Lemma 1
  • proof
  • Lemma 2
  • proof
  • Lemma 3
  • proof
  • Lemma 4
  • proof
  • Lemma 5
  • proof
  • ...and 30 more