Abstract: Sharding technology is one of the core technologies to solve the scalability problem of blockchain.When transactions in the P2P network are aggregated into established shards according to the rules,and the verification nodes are randomly and evenly distributed to each shard,the transactions in this shard may be congested because the transaction verification load of individual shard may far exceed the average load.In order to solve the load imbalance between shards,a shard load balancing method using state reduction is proposed.Firstly,the state reduction model is given,which allows high-performance nodes to store more adjacent states,and the node performance is roughly classified according to this model.Secondly,according to the transaction verification situation in each timeslot,unverified transactions are taken as the remaining load,which is used as the basis for adjusting the verification ability in the next timeslot.Finally,the nodes are scored and graded,and the node selection strategy is given based on the remaining load and the average score of the consensus verification node set.Nodes are allocated reasonably and randomly,and the remaining loads of high load fragments are reduced upward.Experimental verification shows that the shard load balancing method using state reduction effectively balances the unusual load of one shard without reducing the transaction verification rate of a single shard.

Key words: Blockchain, Sharding, State reduction, Load balancing, State synchronization, PBFT