The invention discloses a data storage method and device, and a
server cluster, and relates to the technical field of the data storage. The problem that a user is low in efficiency of reading a
data file in a hyper-
converged storage system since the
data file is stored in a physical
server through a random storage way under the hyper-
converged storage system in the prior art is solved. The methodcomprises the following steps: acquiring access information of the local
data file according to a log file; generating a classified storage rule
bank through the
access frequency to the local data file and the learning on the corresponding data size; and distributing a storage path for the to-be-stored data file carried in a data storage request according to the classified storage rule
bank when accepting the data storage request. The storage method disclosed by the invention is applied to the process of storing the to-be-stored data file in a classification way in the hyper-converged storagesystem.