QScale: Probabilistic Chained Consensus for Moderate-Scale Systems
By: Hasan Heydari, Alysson Bessani, Kartik Nayak
Potential Business Impact:
Makes many computers agree faster and cheaper.
Existing distributed ledger protocols either incur a high communication complexity and are thus suited to systems with a small number of processes (e.g., PBFT), or rely on committee-sampling-based approaches that only work for a very large number of processes (e.g., Algorand). Neither of these lines of work is well-suited for moderate-scale distributed ledgers ranging from a few hundred to a thousand processes, which are common in production (e.g, Redbelly, Sui). The goal of this work is to design a distributed ledger with sub-linear communication complexity per process, sub-quadratic total communication complexity, and low latency for finalizing a block into the ledger, such that it can be used for moderate-scale systems. We propose QScale, a protocol in which every process incurs only $\widetilde{O}(\kappa \sqrt{n})$ communication complexity per-block in expectation, $\widetilde{O}(n\kappa)$ total communication complexity per-block in expectation, and a best-case latency of $O(\kappa)$ rounds while ensuring safety and liveness with overwhelming probability, with $\kappa$ being a small security parameter.
Similar Papers
Near-Optimal Stability for Distributed Transaction Processing in Blockchain Sharding
Distributed, Parallel, and Cluster Computing
Makes blockchain networks handle more transactions safely.
BlueBottle: Fast and Robust Blockchains through Subsystem Specialization
Distributed, Parallel, and Cluster Computing
Makes online money transfers faster and safer.
Exponential Quantum Advantage for Message Complexity in Distributed Algorithms
Quantum Physics
Makes sending messages through networks much faster.