Odontoceti: Ultra-Fast DAG Consensus with Two Round Commitment

TL;DR

Odontoceti presents a practical two-round DAG-based consensus for blockchains, trading a 33% Byzantine fault tolerance for reduced latency in a $n=5f+1$ validator network to achieve a median latency near $0.3$ seconds at $10{,}000$ tx/s. The protocol employs uncertified DAGs, an anchor-based indirect decision rule, and a wave-based two-round structure with early-block production optimization to accelerate commitment, including robust crash fault resilience. Theoretical guarantees are provided for safety and liveness under partial synchrony, and empirical evaluation shows a $20$–$25\%$ latency improvement over the previous state of the art ($Mysticeti$) with similar throughput under realistic network conditions. The work advances practical, high-performance BFT-like consensus for production blockchains and suggests that $5f+1$ networks can be viable with carefully designed two-round protocols and optimization strategies.

Abstract

Users of blockchains value scalability, expecting fast confirmations and immediate transaction processing. Odontoceti, the latest in DAG-based consensus, addresses these concerns by prioritizing low latency and high throughput, making a strategic trade-off in security by operating with a 20% fault tolerance instead of the established 33% level. It is the first DAG-based protocol to achieve commitment in just two communication rounds, delivering median latency of 300 milliseconds while processing 10,000 transactions per second under realistic network conditions. Odontoceti operates with n = 5f + 1 validators and creates an uncertified DAG with a novel decision rule for committing blocks. The protocol includes an optimization that advances progress when participants are slow, benefiting crash fault scenarios which are more common in practice than Byzantine faults. Evaluation results demonstrate 20-25% latency improvements compared to an existing production protocol, validating that reducing wave length from three rounds to two rounds yields meaningful performance benefits. This paper establishes the practical viability of lower fault tolerance consensus protocols for blockchains.

Figures (5)

• Figure 1: An example of linear (left) vs DAG-based (right) block structures. On the left, the red block, which created a fork, will be orphaned as it is not part of the longest chain. On the right, the sub-DAG of the blue block is highlighted.
• Figure 2: DAG Structure. The DAG shows six validators (V1-V6) across six rounds (r1-r6). The colored lines at the bottom indicate waves, where each wave consists of a propose round followed by a decide round. Blocks are colored based on the propose round they were created during. Multiple waves execute in parallel through pipelining, with a new wave beginning each round.
• Figure 3: Walkthrough of the Odontoceti decision rule
• Figure 4: Walkthrough of the early block production optimization
• Figure 5: Graph illustrating Odontoceti's performance under different network and throughput scenarios and in comparison against Mysticeti.

Theorems & Definitions (13)

• LEMMA 1
• LEMMA 2
• LEMMA 3
• COROLLARY 4
• LEMMA 5
• THEOREM 6
• LEMMA 7
• LEMMA 8
• COROLLARY 9
• LEMMA 10