【大数据系列】之Hadoop-MapReduce学习笔记

分布式运算程序

特点：

• 良好扩展性
• 高容错性
• PB级以上的离线处理

缺点：

• 不擅长实时计算
• 不擅长流式计算
• 不擅长DAG(有向无环图)计算

• MrAppMaster:负责整个程序的过程调度及状态调度
• MapTask:负责Map阶段的整个数据处理流程
• ReduceTask:负责Reduce阶段的整个数据处理流程

词频统计

// Map类，继承于org.apache.hadoop.mapreduce.Mapper;
public class WordCountMap extends Mapper {

    Text word = new Text();
    IntWritable value = new IntWritable(1);

    @Override
    protected void map(LongWritable key, Text t, Context context) throws IOException, InterruptedException {
    	// 切分每一行的数据
        String[] words = t.toString().split(" ");
        // 循环输出
        for (String w : words){
            word.set(w);
            context.write(word,value);
        }
    }

}

// Reduce类，继承于org.apache.hadoop.mapreduce.Reducer
public class WordCountReduce extends Reducer {

    int sum;
    IntWritable count = new IntWritable();

    @Override
    protected void reduce(Text key, Iterable values, Context context) throws IOException, InterruptedException {

        sum = 0;
		// 累加求和
        for (IntWritable v: values){
            sum += v.get();
        }
        count.set(sum);
        context.write(key,count);

    }
}

public class WordCountDriver {

    public static void main(String[] args) throws Exception {
    	// 设置操作的用户
        System.setProperty("HADOOP_USER_NAME","root");

        Configuration conf = new Configuration();
        Job job = Job.getInstance(conf);
		// 配置Driver类
        job.setJarByClass(WordCountDriver.class);
		// 配置Map,Reduce类
        job.setMapperClass(WordCountMap.class);
        job.setReducerClass(WordCountReduce.class);
		// 配置Map的输出Key类型及Value类型
        job.setMapOutputKeyClass(Text.class);
        job.setMapOutputValueClass(IntWritable.class);
		// 配置Reduce的输出Key类型及Value类型
        job.setOutputKeyClass(Text.class);
        job.setOutputValueClass(IntWritable.class);
		// 设置输入路径以及输出路径
        FileInputFormat.setInputPaths(job,new Path("hdfs://192.168.19.16:9000/mol/test.txt"));
        FileOutputFormat.setOutputPath(job,new Path("hdfs://192.168.19.16:9000/mol/wordcount"));
		// 启动任务
        Boolean result = job.waitForCompletion(true);
        System.exit(result ? 0:1);

    }
}