Abstract: A method and system for reducing replication factor in a file system are provided. In some embodiments, two or more requested files may be grouped together under a leaf directory for RAID process. All data under the directory are grouped into one or more data stripes, each of which comprises a plurality of data blocks. One or more parity data blocks may be generated for each data stripe according to a computing algorithm, such as an exclusive OR (XOR) code or a Reed-Solomon (RS) code. Parity blocks corresponding to the one or more data stripes are concatenated into one parity file. Data blocks of the two or more requested files and their corresponding parity blocks are stored in separate partitions and/or separate storage drives of the file system.