Disclosed is a cold and hot area dividing energy-saving storage method based on cluster node load state prediction, belonging to the field of technical processing of information. The concept of the method is introduced first in the present application, then the execution steps of the method are described, and a pseudocode of the method is realized, next the energy-saving effect of an energy-saving policy is theoretically analyzed, then the procedure of an experiment is introduced, wherein the procedure comprises an environment of the experiment, selection of a data set and an access format of a data block, and finally a result of the experiment is analyzed. By means of quantitative analysis and comparison with energy consumption, in the case where cold and hot areas are not divided, in a Hadoop cluster, it is verified that an energy-saving storage policy of the present application has a certain actual energy-saving effect.
展开▼
