The expansion of blockchain storage has become a major problem limiting the application of blockchain. From the perspective of improving the scalability of blockchain storage, a DHT (distributed hash table)-based blockchain dual-sharding storage extension mechanism (DBDSM) is proposed. The nodes in the network are divided into m DHT clusters. Each cluster includes n nodes, and stores 1/m of the transaction data, and the nodes within each cluster store part of data allocated to that cluster. In this way, node storage pressure is alleviated. Furthermore, a hybrid query mechanism has been designed to achieve efficient querying of transaction data, without changing the original state data query. Simulation results showed without changing the original state data query, that the storage space consumed by the nodes is only s/(m × n) of that used in the traditional method; when the number of faulty nodes in the cluster does not exceed s − 1, the integrity of blockchain data can still be ensured. For transaction data queries, the average number of hops was 1.99, greatly improving query efficiency in the sharded state.

Currently, blockchain is facing a serious storage explosion problem. While most storage scaling schemes are focused on permissionless blockchain, we propose the ring-overlap mechanism for consortium blockchain and use it to scale Hyperledger Fabric. In our scheme, all accounting nodes are divided into clusters, and each cluster contains several nodes; then, a portion of the entire block data is stored in a cluster. Block data is stored overlappingly on some cluster nodes, and each block is guaranteed to have some copies in a cluster. Theoretical analysis and simulation show that the storage occupied by nodes is significantly reduced in blockchain applications with frequent transactions, and the mechanism can still guarantee data integrity in the case of partial node failures in a single cluster. Furthermore, for transaction-frequent applications, storage space consumption can be significantly reduced without increasing excessive query time overhead.