Moonshot: Optimizing Chain-Based Rotating Leader BFT via Optimistic Proposals Isaac Doidge, Raghavendra Ramesh, Nibesh Shrestha, and Joshua Tobkin Supra Research Abstract—Existing chain-based rotating-leader BFT SMR protocols for the partially synchronous network model with constant commit latencies incur block periods of at least 2δ (where δ is the message transmission latency). While a protocol with a block period of δ exists under the synchronous model, its commit latency is linear in the size of the system. To close this gap, we present the first chain-based BFT SMR protocols with δ delay between the proposals of consecutive honest leaders and commit latencies of 3δ. We present three protocols for the partially synchronous model under different notions of optimistic responsiveness, two of which implement pipelining. All of our protocols achieve reorg resilience and two have short view lengths; properties that many existing chain-based BFT SMR protocols lack. We present an evaluation of our protocols in a wide-area network wherein they demonstrate significant increases in throughput and reductions in latency compared to the state-of-the-art, Jolteon. Our results also demonstrate that techniques commonly employed to reduce communication complexity—such as vote-pipelining and the use of designated vote-aggregators—actually reduce practical performance in many settings. ### I. INTRODUCTION Blockchain networks have become increasingly popular as mechanisms for facilitating decentralised, immutable and verifiable computation and storage. These networks leverage Byzantine fault-tolerant (BFT) consensus protocols to ensure that their participants (called nodes) execute the same sequence of operations (called transactions), despite some of them exhibiting arbitrary failures. Many blockchain networks also prioritize fairness; i.e., they strive to ensure that i) client transactions are processed promptly, without granting any client an unfair advantage over the others, and ii) nodes have an equal opportunity to be rewarded for the work that they do in the system. Public blockchain networks in particular also tend to be large, supporting hundreds (e.g. [26]) or thousands (e.g. [6]) of nodes in the pursuit of decentralization, and aim to cater to many concurrent clients. Accordingly, the consensus protocols driving these networks need to be efficient, maximising transaction throughput and minimising end-to-end commit latency (i.e., the time between a client submitting a transaction and it being executed by the blockchain). To these ends, prior works [32], [37], [20], [4], [11], [21], [27] have leveraged two key strategies: i) block chaining, and; ii) frequent leader rotation. In the block chaining (or chained) paradigm, transactions are grouped into blocks that explicitly reference one or more existing blocks (called the parents of the block), typically by including their hashes. This enables an optimization called pipelining, wherein the voteMoonshot: Optimizing Chain-Based Rotating Leader BFT via Optimistic Proposals Isaac Doidge, Raghavendra Ramesh, Nibesh Shrestha, and Joshua Tobkin Supra Research Abstract—Existing chain-based rotating-leader BFT SMR protocols for the partially synchronous network model with constant commit latencies incur block periods of at least 2δ (where δ is the message transmission latency). While a protocol with a block period of δ exists under the synchronous model, its commit latency is linear in the size of the system. To close this gap, we present the first chain-based BFT SMR protocols with δ delay between the proposals of consecutive honest leaders and commit latencies of 3δ. We present three protocols for the partially synchronous model under different notions of optimistic responsiveness, two of which implement pipelining. All of our protocols achieve reorg resilience and two have short view lengths; properties that many existing chain-based BFT SMR protocols lack. We present an evaluation of our protocols in a wide-area network wherein they demonstrate significant increases in throughput and reductions in latency compared to the state-of-the-art, Jolteon. Our results also demonstrate that techniques commonly employed to reduce communication complexity—such as vote-pipelining and the use of designated vote-aggregators—actually reduce practical performance in many settings. ### I. INTRODUCTION Blockchain networks have become increasingly popular as mechanisms for facilitating decentralised, immutable and verifiable computation and storage. These networks leverage Byzantine fault-tolerant (BFT) consensus protocols to ensure that their participants (called nodes) execute the same sequence of operations (called transactions), despite some of them exhibiting arbitrary failures. Many blockchain networks also prioritize fairness; i.e., they strive to ensure that i) client transactions are processed promptly, without granting any client an unfair advantage over the others, and ii) nodes have an equal opportunity to be rewarded for the work that they do in the system. Public blockchain networks in particular also tend to be large, supporting hundreds (e.g. [26]) or thousands (e.g. [6]) of nodes in the pursuit of decentralization, and aim to cater to many concurrent clients. Accordingly, the consensus protocols driving these networks need to be efficient, maximising transaction throughput and minimising end-to-end commit latency (i.e., the time between a client submitting a transaction and it being executed by the blockchain). To these ends, prior works [32], [37], [20], [4], [11], [21], [27] have leveraged two key strategies: i) block chaining, and; ii) frequent leader rotation. In the block chaining (or chained) paradigm, transactions are grouped into blocks that explicitly reference one or more existing blocks (called the parents of the block), typically by including their hashes. This enables an optimization called pipelining, wherein the vote