Constructing a Fine-Grained Mapping Algorithm for Blocks and Rack-Mounted Storage Nodes

In a heterogeneous encoding environment, cloud storage systems face a series of challenges caused by frequent data repairs, including the imbalance of erasure coding data transmission, the uneven read load of data racks and nodes, and the high overhead of data repair that results from these issues. This paper aims to address the key problems in cloud storage systems based on erasure coding technology: how to reduce storage costs while ensuring data reliability, and improve the overall system performance and high fault tolerance of the data. This study first proposes a Fine-Grained Mapping Algorithm (FEM) for blocks and storage nodes on racks, which can achieve precise mapping between blocks and storage nodes. The aim is to balance the load of storage nodes and racks simultaneously, accelerate the data repair process, avoid load imbalance caused by the repair program reading different numbers of blocks from different storage nodes, balance the network resource differences between storage nodes, and to some extent, ensure the security of data storage.