Blockchain is the foundation of emerging applications such as smart contracts, Non-fungible token (NFT) and metaverse. A key issue is that blockchain requires massive storage space, which limits its deployment in resource limited end devices, e.g., Internet of Things. Recently, coded blockchain is proposed to reduce the storage requirement of blockchain whilst guaranteeing its security and data integrity. Coded blockchain encodes blocks into coded symbols, which are then distributively stored by clients. A key challenge when applying coded blockchain in resource restricted networks is to ensure all clients store the same, and also the minimum, number of coded blocks. To this end, this paper addresses a novel problem that minimizes the maximum (min-max) storage requirement of clients. It formulates the said problem as an integer linear program (ILP). It then proposes centralized algorithms to improve the computational efficiency of storage assignments. Moreover, it presents distributed algorithms that satisfy the distributive property of blockchain. Numerical results show that the proposed distributed algorithm with a short length code reduces the min-max storage of clients by 80% compared with traditional blockchain. In addition, the computational complexity of distributed algorithms is significantly lower than centralized algorithms.