A big-data processing method is provided. The method includes obtaining multiple key-value pair sets, where each of the key-value pair sets includes at least one key-value pair, and remainders of modulo operations on all key-value pairs in each key-value pair set are the same, calculating a sum of all values included in each key-value pair set to obtain a total value of the values included in each key-value pair set, obtaining a load status of each reduce processor in a reduce processor set, assigning a corresponding key-value pair set to each reduce processor according to the total value of the values included in each key-value pair set and the load status of each reduce processor, and processing the assigned key-value pair set using each reduce processor.