Sharding SMR with Optimal-size Shards for Highly Scalable Blockchains
Primary Investigator:
Aniket Kate
Jianting Zhang, Zhongtang Luo, Raghavendra Ramesh, Aniket Kate
Abstract
Blockchain relies on State Machine Replication (SMR) to enable trustless nodes to uphold a consistent ledger while tolerating Byzantine faults. With the rapid growth of decentralized web3 platforms and applications, a central challenge of blockchain systems is scalability, which can be evaluated with two metrics: high performance and large network. However, existing blockchain systems struggle to simultaneously achieve both scalability metrics while requiring to guarantee the underlying security properties of SMR—safety and liveness. In this poster, we present a novel blockchain architecture addressing this dilemma by sharding the SMR. Our architecture builds upon two core insights: ordering-processing sharding scheme and safety-liveness separation. Specifically, the ordering-processing sharding scheme securely accommodates a large number of nodes by dividing them into multiple shards, enhancing the network scale. Additionally, the safety-liveness separation allows each shard to consider the security properties of SMR against Byzantine failures separately, by which the system can create more optimal-size shards to process transactions in parallel, enhancing performance. The preliminary experiments show the efficacy of our architecture in scaling blockchains.