1.Partitioner是partitioner的基类，如果需要定制Partitioner也需要继承该类。

 

2. HashPartitioner是mapreduce的默认partitioner。计算方法是 which reducer=(key.hashCode() & Integer.MAX_VALUE) % numReduceTasks，得到当前的目的reducer。

 

3.说明，Partitioner是在Mapper执行完成，Reducer执行前。它有两个参数，就是Mapper的输出参数，在这里，有几个Reducer就有几个Partitioner

 

4.根据数据分区，将数据传入不同的Reducer中

 

示例：

public static class ProviderPartitioner extends Partitioner<Text, DataBean>

            private static Map<String, Integer> providerMap = new HashMap<String, Integer>();

            static {

            public int getPartition(Text key, DataBean value, int numPartitions) {

                  String accountString = key.toString();

                  String sub_accString = accountString.substring(0, 3);

                  Integer codeInteger =  providerMap.get(sub_accString);

                  if (codeInteger == null)

                        codeInteger = 0;

                  return codeInteger;