Raptr: The Speedy, Resilient BFT Protocol That Bridges Optimism and Pessimism đđť
Raptr is a new Byzantine Fault-Tolerant (BFT) state machine replication protocol designed for blockchain systems, aiming to deliver high throughput, low latency, and robustness. Unlike traditional leader-based BFT protocols, which struggle with throughput due to a single leaderâs bandwidth limitations, or DAG-based protocols, which can falter under network issues, Raptr introduces a novel Prefix Consensus mechanism. This allows replicas to vote on partial blocks (prefixes) rather than requiring full data availability, enabling progress even when some data is missing.
Key highlights:
⢠Performance: In tests with 100 geo-distributed nodes, Raptr achieves up to 260,000 transactions per second (TPS) with sub-second latency (610ms at 10,000 TPS, 755ms at 250,000 TPS).
⢠Robustness: It handles network glitches (e.g., 1% message loss) with minimal degradation, unlike other protocols that see sharp performance drops.
⢠Latency: Raptr cuts latency to 5 message delays in the common case, close to the theoretical minimum, by integrating optimistic data dissemination with background certification.
⢠Innovations: Combines the low latency of leader-based protocols (like Jolteon) with the high throughput of DAG-based systems, using prefix voting to avoid stalling on missing data.
The protocol builds on the #Aptos blockchainâs Jolteon and Quorum Store, enhancing them with Baby Raptr (an optimistic variant) and Raptrâs prefix voting to balance speed and resilience. Evaluations show it outperforms state-of-the-art protocols like Shoal++ and Mysticeti, especially under adverse conditions.
Compared to DAG-based protocols like Mysticeti, Raptr feels more practical for real-world blockchain use where networks arenât always perfect. If youâre into blockchain tech, Raptrâs worth a lookâit could be a game-changer for scalable, robust systems!
If you enjoy my content, feel free to follow me â¤ď¸
