Content similarity-based and distributed storage replica replacement algorithm

The invention discloses a content similarity-based and distributed storage replica replacement algorithm, and the algorithm is characterized by calculating a replica value based on the content similarity first, then calculating a replication cost, and finally considering prediction elements and cost elements comprehensively for putting forward the content similarity-based and distributed storage replica replacement algorithm. Owing to differences of replica sizes and connection bandwidths between computing nodes and storage nodes, the replicas have different elimination costs. The content similarity-based and distributed storage replica replacement algorithm extracts the least frequently used replicas through the content similarity-based replica replacement algorithm, finds least recently used replicas through the distributed storage replica replacement algorithm and then realizes replica displacement through comprehensive use of the two algorithms, thereby improving average working efficiency, reducing consumption of network resources and being significant to the field of the distributed storage.