The invention relates to a distributed cluster-based data processing method and system. In all backup data of all data blocks under the same data sub-region, the data sub-region is taken as a unit, first backup data are stored in a data node where the same is located, and all the other backup data are respectively stored in two data nodes which are in other data nodes, have fewest data sub-regions, and are respectively described as first and second backup nodes; and the data sub-region is transferred into a data node with fewer data sub-regions when a data sub-region server crashes or does notprovide service. After a node fault occurs and sub-region transfer is caused, a sub-region server providing service for the sub-region after transfer still acquires data locally without acquiring data from another node through a network, an effect that the data in the data sub-region can still be accessed without adopting the network is realized, access efficiency of distributed databases after the sub-region server node fault is improved, and the problem that except for a main compression method, means of increasing distributed-database localization rates are lacked is solved.