In this paper, according to the resource management problems brought by a large number of replicas, a multi-replica clustering management method based on limited-coding is proposed. In this method, according to the process of creating new replicas from existent single replica, replicas are partitioned into different hierarchies and clusters. Then replicas are coded and managed based on the user-defined limited-coding rule consisting of replica hierarchy and replica sequence, which can also dispose the alteration of clusters caused by dynamic adjustments on replicas (replica addition or replica removal) effectively. After that, a management model of centralization in local and peer to peer in wide area is adopted to organize replicas, and the cost of reconciling consistency can be greatly depressed combining with defined minimal-time of update propagation. The relevance between the coding rule and the number of replicas, and the solutions to replica failure and replica recover are discussed. The results of the performance evaluation show that the clustering method is an efficient way to manage a large number of replicas, achieving good scalability, not sensitive to moderate node failure, and adapting well to applications with frequent updates.