The invention discloses a data distributed storage system and method. The system comprises a node cluster module, a data import module and a storage module. The node cluster module is used for connecting data nodes in a cluster with corresponding management nodes; the data import module is used for scanning input data according to data blocks with sizes set and loading the input data into a memory, data in the memory are grouped according to characteristic values of the data, and the grouped data are sent to the corresponding data nodes; the storage module is used for storing data fragmentations in the memory after the data nodes receive file fragmentations, and the data nodes output logs to a hard disk; whether the data in the memory exceed a set threshold value or not is judged, if the data in the memory exceed the set threshold value, the data are reorganized and compressed and then written into the hard disk, and corresponding log files restored through user memory data are deleted. According to the system and method, the cluster based on memory computing power can be accelerated; real-time loading and processing capacity for large-scale data can be improved, and response time of the system is shortened.