Blockchain ensures data security and consistency without relying on trusted third-party institutions through full replication mechanisms and consensus protocols. However, this design significantly limits system scalability. Sharding addresses this limitation by partitioning the blockchain into multiple parallel working shards, with state sharding further distributing state data across these shards, significantly improving system throughput and storage efficiency. Currently, state sharding faces several challenges, including the following three problems: (1) Cross-shard transactions require collaborative processing across multiple shards, introducing additional communication and consensus overhead, necessitating improvements in processing efficiency; (2) State partitioning optimization is needed to reduce the proportion of cross-shard transactions and balance shard workloads; (3) Efficient state data migration between shards is required during dynamic state partitioning and adjustment. This study systematically reviews the key challenges and mainstream implementations of state sharding, focusing on cross-shard transaction processing, state allocation, and state migration. Finally, the study identifies current limitations of state sharding and suggests possible directions for future research.