Blockchain technology, despite its growing importance, has been argued for its poor scalability. Sharding has become an effective approach to addressing the scalability issue faced in the blockchain. It divides nodes into multiple groups (shards) so that they can process transactions in parallel. However, the efficiency and security in existing blockchain sharding systems, despite being the basic design objectives, still suffer from many problems. To address these issues and achieve an efficient and secure blockchain sharding system, this thesis makes the following contributions.

First, existing blockchain sharding protocols do not scale well when tackling smart contracts (functions executing various logic). We present Jenga, a novel sharding-based approach for efficient smart contract...[ Read more ]

Blockchain technology, despite its growing importance, has been argued for its poor scalability. Sharding has become an effective approach to addressing the scalability issue faced in the blockchain. It divides nodes into multiple groups (shards) so that they can process transactions in parallel. However, the efficiency and security in existing blockchain sharding systems, despite being the basic design objectives, still suffer from many problems. To address these issues and achieve an efficient and secure blockchain sharding system, this thesis makes the following contributions.

First, existing blockchain sharding protocols do not scale well when tackling smart contracts (functions executing various logic). We present Jenga, a novel sharding-based approach for efficient smart contract processing. It requires all shards to share the logic for all contracts, hence multiple smart contracts can be executed within one round. Moreover, different shards store distinct states (state shards), and several orthogonal execution channels are established based on the state shards to reduce cross-shard communication. We implement Jenga and evaluation results illustrate that it achieves more than 1.5x throughput gain compared to the existing cutting-edge work.

Second, existing works, for security, tend to configure large shard sizes to maintain a small fraction of intra-shard malicious nodes, at an expense of limited concurrency. This leads to limited concurrency. Therefore, we propose CoChain, a highly concurrent and recoverable blockchain sharding system. It allows shards to have a larger fraction of malicious nodes (up to 2/3), thus reducing the shard size and increasing concurrency. To ensure security, CoChain requires shards to monitor each other, hence a shard with a high fraction of malicious nodes can recover from failure with the help of other shards. We implement CoChain based on Harmony and experimental results show that CoChain achieves a 35x throughput gain compared with Harmony.

Third, existing blockchain sharding, during intra-shard consensus and cross-shard transaction processing, either faces security issues or sacrifices great performance for security. Therefore, we propose SS-Chain for enhanced intra/cross-shard security and performance. We design pipelined consensus with less communication overhead to improve intra-shard consensus efficiency and defend against attacks via frequent leader rotations. We attach proofs to cross-shard transactions to ensure cross-shard security and improve efficiency through transaction batching and proof pruning. We implement SS-Chain and large-scale evaluation results demonstrate that it achieves a throughput of more than 10,000 tx/sec even under malicious behaviors