A Dynamic Sharding Scheme For Blockchain Based On Graph Partitioning

Blockchain is a decentralized distributed ledger that is stored in a block chain, cannot be tampered with, is safe and trustworthy with the help of cryptography and consensus mechanism. However, the current performance and scalability of blockchain hinders its practical application in wider areas, so blockchain sharding technology was proposed. Sharding technology divides the blockchain into multiple shards. Each shard is responsible for processing a part of transactions and status. Multiple shards verify and process transactions in parallel, which improves the performance of the blockchain. Although sharding technology has potential in theory, in practical applications, it still faces a series of problems, such as shard load imbalance and a high proportion of cross-shard transactions. In order to solve the problems of unbalanced shard load and high proportion of cross-shard transactions in the blockchain sharding system, the DLDG (Dynamic Linear Deterministic Greedy) dynamic account division algorithm is proposed. DLDG constructs the shard load and cross-shard transaction ratio. The shard transaction rate balancing model periodically divides the account transaction network graph through historical transaction data, thereby achieving shard load balancing and reducing the cross-shard transaction rate. Experimental results show that this scheme alleviates performance problems caused by load imbalance and high cross-shard transaction rates.