如何使用Hadoop读写数据库

在我们的一些应用程序中，常常避免不了要与数据库进行交互，而在我们的hadoop中，有时候也需要和数据库进行交互，比如说，数据分析的结果存入数据库，或者是，读取数据库的信息写入HDFS上，不过直接使用MapReduce操作数据库，这种情况在现实开发还是比较少，一般我们会采用Sqoop来进行数据的迁入，迁出，使用Hive分析数据集，大多数情况下，直接使用Hadoop访问关系型数据库，可能产生比较大的数据访问压力，尤其是在数据库还是单机的情况下，情况可能更加糟糕，在集群的模式下压力会相对少一些。 

那么，今天散仙就来看下，如何直接使用Hadoop1.2.0的MR来读写操作数据库，hadoop的API提供了DBOutputFormat和DBInputFormat这两个类，来进行与数据库交互，除此之外，我们还需要定义一个类似JAVA Bean的实体类，来与数据库的每行记录进行对应，通常这个类要实现Writable和DBWritable接口，来重写里面的4个方法以对应获取每行记录里面的各个字段信息。 

下面，我们先来看下如何使用MR来读取数据库的数据，并写入HDFS上， 
数据表的截图如下所示， 

 

实体类定义代码： 

<pre name="code" class="java">package com.qin.operadb; 

import java.io.DataInput; 

import java.io.DataOutput; 

import java.io.IOException; 

import java.sql.PreparedStatement; 

import java.sql.ResultSet; 

import java.sql.SQLException; 

import org.apache.hadoop.io.Text; 

import org.apache.hadoop.io.Writable; 

import org.apache.hadoop.mapreduce.lib.db.DBWritable; 

/*** 

* 封装数据库实体信息 

* 的记录 

* 

* 搜索大数据技术交流群：376932160 

* 

* **/ 

public class PersonRecoder implements Writable,DBWritable { 

public int id;//对应数据库中id字段 

public String name;//对应数据库中的name字段 

public int age;//对应数据库中的age字段 

@Override 

public void readFields(ResultSet result) throws SQLException { 

this.id=result.getInt(1); 

this.name=result.getString(2); 

this.age=result.getInt(3); 

} 

@Override 

public void write(PreparedStatement stmt) throws SQLException { 

stmt.setInt(1, id); 

stmt.setString(2, name); 

stmt.setInt(3, age); 

} 

@Override 

public void readFields(DataInput arg0) throws IOException { 

// TODO Auto-generated method stub 

this.id=arg0.readInt(); 

this.name=Text.readString(arg0); 

this.age=arg0.readInt(); 

} 

@Override 

public void write(DataOutput out) throws IOException { 

// TODO Auto-generated method stub 

out.writeInt(id); 

Text.writeString(out, this.name); 

out.writeInt(this.age); 

} 

@Override 

public String toString() { 

// TODO Auto-generated method stub 

return "id: "+id+"  年龄: "+age+"   名字："+name; 

} 

} 

</pre> 

MR类的定义代码，注意是一个Map Only作业： 

<pre name="code" class="java">package com.qin.operadb; 

import java.io.IOException; 

import org.apache.hadoop.conf.Configuration; 

import org.apache.hadoop.fs.FileSystem; 

import org.apache.hadoop.fs.Path; 

import org.apache.hadoop.io.LongWritable; 

import org.apache.hadoop.io.Text; 

import org.apache.hadoop.mapred.JobConf; 

import org.apache.hadoop.mapred.lib.IdentityReducer; 

import org.apache.hadoop.mapreduce.Job; 

import org.apache.hadoop.mapreduce.Mapper; 

import org.apache.hadoop.mapreduce.lib.db.DBConfiguration; 

import org.apache.hadoop.mapreduce.lib.db.DBInputFormat; 

import org.apache.hadoop.mapreduce.lib.output.FileOutputFormat; 

public class ReadMapDB { 

/** 

* Map作业读取数据记录数 

* 

* **/ 

private static class DBMap extends Mapper<LongWritable, PersonRecoder , LongWritable, Text>{ 

@Override 

protected void map(LongWritable key, PersonRecoder value,Context context) 

throws IOException, InterruptedException { 

context.write(new LongWritable(value.id), new Text(value.toString())); 

} 

} 

public static void main(String[] args)throws Exception { 

JobConf conf=new JobConf(ReadMapDB.class); 

//Configuration conf=new Configuration(); 

   // conf.set("mapred.job.tracker","192.168.75.130:9001"); 

//读取person中的数据字段 

// conf.setJar("tt.jar"); 

//注意这行代码放在最前面，进行初始化，否则会报 

DBConfiguration.configureDB(conf, "com.mysql.jdbc.Driver", "jdbc:mysql://192.168.211.36:3306/test", "root", "qin"); 

/**要读取的字段信息**/ 

String fileds[]=new String[]{"id","name","age"}; 

/**Job任务**/ 

Job job=new Job(conf, "readDB"); 

System.out.println("模式：  "+conf.get("mapred.job.tracker")); 

/**设置数据库输入格式的一些信息**/ 

DBInputFormat.setInput(job, PersonRecoder.class, "person", null, "id", fileds); 

/***设置输入格式*/ 

job.setInputFormatClass(DBInputFormat.class); 

job.setOutputKeyClass(LongWritable.class); 

job.setOutputValueClass(Text.class); 

job.setMapperClass(DBMap.class); 

String path="hdfs://192.168.75.130:9000/root/outputdb"; 

FileSystem fs=FileSystem.get(conf); 

Path p=new Path(path); 

if(fs.exists(p)){ 

fs.delete(p, true); 

System.out.println("输出路径存在，已删除！"); 

} 

FileOutputFormat.setOutputPath(job,p ); 

System.exit(job.waitForCompletion(true) ? 0 : 1);  

} 

} 

</pre> 

写入到HDFS目录下数据集： 

 

读取相对比较简单，需要注意的第一注意JDBC的驱动jar包要在各个节点上分别上传一份，第二是在main方法里的驱动类的编写顺序，以及数据信息的完整，才是正确连接数据库并读取的关键。 

下面来看下，如何使用MR，分析完数据后的结果，写入在数据库中，散仙本篇测试的是一个简单的WordCount的统计。我们先来看下数据库表的信息： 

 

实体类定义代码： 

<pre name="code" class="java">package com.qin.operadb; 

import java.io.DataInput; 

import java.io.DataOutput; 

import java.io.IOException; 

import java.sql.PreparedStatement; 

import java.sql.ResultSet; 

import java.sql.SQLException; 

import org.apache.hadoop.io.Text; 

import org.apache.hadoop.io.Writable; 

import org.apache.hadoop.mapreduce.lib.db.DBWritable; 

public class WordRecoder implements Writable,DBWritable { 

public String word; 

public int count; 

@Override 

public void readFields(ResultSet rs) throws SQLException { 

this.word=rs.getString(1); 

this.count=rs.getInt(2); 

} 

@Override 

public void write(PreparedStatement ps) throws SQLException { 

ps.setString(1, this.word); 

ps.setInt(2, this.count); 

} 

@Override 

public void readFields(DataInput in) throws IOException { 

  this.word=Text.readString(in); 

  this.count=in.readInt(); 

} 

@Override 

public void write(DataOutput out) throws IOException { 

  

   

Text.writeString(out, this.word); 

out.writeInt(count); 

} 

} 

</pre> 

统计的2个文件的内容所示： 

 

MR的核心类代码： 

<pre name="code" class="java">package com.qin.operadb; 

import java.io.IOException; 

import java.util.StringTokenizer; 

import org.apache.hadoop.fs.FileSystem; 

import org.apache.hadoop.fs.Path; 

import org.apache.hadoop.io.IntWritable; 

import org.apache.hadoop.io.LongWritable; 

import org.apache.hadoop.io.Text; 

import org.apache.hadoop.mapred.JobConf; 

import org.apache.hadoop.mapreduce.Job; 

import org.apache.hadoop.mapreduce.Mapper; 

import org.apache.hadoop.mapreduce.Reducer; 

import org.apache.hadoop.mapreduce.lib.db.DBConfiguration; 

import org.apache.hadoop.mapreduce.lib.db.DBInputFormat; 

import org.apache.hadoop.mapreduce.lib.db.DBOutputFormat; 

import org.apache.hadoop.mapreduce.lib.input.FileInputFormat; 

import org.apache.hadoop.mapreduce.lib.input.TextInputFormat; 

import org.apache.hadoop.mapreduce.lib.output.FileOutputFormat; 

public class WriteMapDB { 

private static class WMap extends Mapper<LongWritable, Text, Text, IntWritable>{ 

/*** 

* Mapper的参数类型介绍 

* K,V,K,V分别依次代表 

* Map作业输入类型的K，输入类型的V 

* 后面两个是输出类型的K，输出类型的V 

* 后面的两个与 context.write(word, one); 

* 的两个参数是对应的 

* **/ 

private Text word=new Text(); 

private IntWritable one=new IntWritable(1); 

@Override 

protected void map(LongWritable key, Text value,Context context) 

throws IOException, InterruptedException { 

String line=value.toString(); 

//处理记事本UTF-8的BOM问题 

            if (line.getBytes().length > 0) { 

                if ((int) line.charAt(0) == 65279) { 

                    line = line.substring(1); 

                } 

            } 

StringTokenizer st=new StringTokenizer(line); 

while(st.hasMoreTokens()){ 

word.set(st.nextToken());//设置单词 

context.write(word, one); 

} 

} 

} 

/*** 

* 由于在reduce中，需要向数据库里写入 

* 数据，所以跟combine，不能共用 

* 

* 

* 

* ***/ 

private static class WCombine extends Reducer<Text, IntWritable, Text, IntWritable>{ 

@Override 

protected void reduce(Text text, Iterable<IntWritable> value,Context context) 

throws IOException, InterruptedException { 

int sum=0; 

for(IntWritable iw:value){ 

sum+=iw.get(); 

} 

context.write(text, new IntWritable(sum)); 

} 

} 

/** 

* 

* Reduce类 

* 

* **/ 

private static class WReduce extends Reducer<Text, IntWritable, WordRecoder, Text>{ 

@Override 

protected void reduce(Text key, Iterable<IntWritable> values,Context context) 

throws IOException, InterruptedException { 

  int sum=0; 

  for(IntWritable s:values){ 

  sum+=s.get(); 

  } 

  

  

  WordRecoder wr=new WordRecoder(); 

  wr.word=key.toString(); 

  wr.count=sum; 

  

  //写出到数据库里 

  context.write(wr, new Text()); 

} 

} 

public static void main(String[] args)throws Exception { 

JobConf conf=new JobConf(WriteMapDB.class); 

//Configuration conf=new Configuration(); 

   // conf.set("mapred.job.tracker","192.168.75.130:9001"); 

//读取person中的数据字段 

  //conf.setJar("tt.jar"); 

// conf.setNumReduceTasks(1); 

//注意这行代码放在最前面，进行初始化，否则会报 

/**建立数据库连接**/ 

DBConfiguration.configureDB(conf, "com.mysql.jdbc.Driver", "jdbc:mysql://192.168.211.36:3306/test?characterEncoding=utf-8", "root", "qin"); 

String fileds[]=new String[]{"word","count"}; 

Job job=new Job(conf, "writeDB"); 

System.out.println("运行模式：  "+conf.get("mapred.job.tracker")); 

/**设置输出表的的信息  第一个参数是job任务，第二个参数是表名，第三个参数字段项**/ 

DBOutputFormat.setOutput(job, "wordresult", fileds); 

/**设置DB的输入路径**/ 

job.setInputFormatClass(TextInputFormat.class); 

/**设置DB的输出路径**/ 

job.setOutputFormatClass(DBOutputFormat.class); 

/***设置Reduce的个数为1，可以得到全局统计的数字 

* 但，需要注意，在分布式环境下，最好不要设置为1，Reduce的个数 

* 正是Hadoop并发能力的体现 

* 

* **/ 

// job.setNumReduceTasks(1); 

/**设置输出K路径**/ 

job.setOutputKeyClass(Text.class); 

/**设置输出V路径**/ 

job.setOutputValueClass(IntWritable.class); 

/**设置Map类**/ 

job.setMapperClass(WMap.class); 

/**设置Combiner类**/ 

job.setCombinerClass(WCombine.class); 

/**设置Reduce类**/ 

job.setReducerClass(WReduce.class); 

/**设置输入路径*/ 

FileInputFormat.setInputPaths(job, new Path("hdfs://192.168.75.130:9000/root/input")); 

System.exit(job.waitForCompletion(true) ? 0 : 1);  

} 

} 

</pre> 

运行状态如下所示： 

<pre name="code" class="java">运行模式：  192.168.75.130:9001 

14/03/26 20:26:59 WARN mapred.JobClient: Use GenericOptionsParser for parsing the arguments. Applications should implement Tool for the same. 

14/03/26 20:27:01 INFO input.FileInputFormat: Total input paths to process : 2 

14/03/26 20:27:01 WARN util.NativeCodeLoader: Unable to load native-hadoop library for your platform... using builtin-java classes where applicable 

14/03/26 20:27:01 WARN snappy.LoadSnappy: Snappy native library not loaded 

14/03/26 20:27:01 INFO mapred.JobClient: Running job: job_201403262328_0006 

14/03/26 20:27:02 INFO mapred.JobClient:  map 0% reduce 0% 

14/03/26 20:27:10 INFO mapred.JobClient:  map 50% reduce 0% 

14/03/26 20:27:11 INFO mapred.JobClient:  map 100% reduce 0% 

14/03/26 20:27:18 INFO mapred.JobClient:  map 100% reduce 33% 

14/03/26 20:27:19 INFO mapred.JobClient:  map 100% reduce 100% 

14/03/26 20:27:20 INFO mapred.JobClient: Job complete: job_201403262328_0006 

14/03/26 20:27:20 INFO mapred.JobClient: Counters: 28 

14/03/26 20:27:20 INFO mapred.JobClient:   Job Counters 

14/03/26 20:27:20 INFO mapred.JobClient:     Launched reduce tasks=1 

14/03/26 20:27:20 INFO mapred.JobClient:     SLOTS_MILLIS_MAPS=10345 

14/03/26 20:27:20 INFO mapred.JobClient:     Total time spent by all reduces waiting after reserving slots (ms)=0 

14/03/26 20:27:20 INFO mapred.JobClient:     Total time spent by all maps waiting after reserving slots (ms)=0 

14/03/26 20:27:20 INFO mapred.JobClient:     Launched map tasks=2 

14/03/26 20:27:20 INFO mapred.JobClient:     Data-local map tasks=2 

14/03/26 20:27:20 INFO mapred.JobClient:     SLOTS_MILLIS_REDUCES=8911 

14/03/26 20:27:20 INFO mapred.JobClient:   File Output Format Counters 

14/03/26 20:27:20 INFO mapred.JobClient:     Bytes Written=0 

14/03/26 20:27:20 INFO mapred.JobClient:   FileSystemCounters 

14/03/26 20:27:20 INFO mapred.JobClient:     FILE_BYTES_READ=158 

14/03/26 20:27:20 INFO mapred.JobClient:     HDFS_BYTES_READ=325 

14/03/26 20:27:20 INFO mapred.JobClient:     FILE_BYTES_WRITTEN=182065 

14/03/26 20:27:20 INFO mapred.JobClient:   File Input Format Counters 

14/03/26 20:27:20 INFO mapred.JobClient:     Bytes Read=107 

14/03/26 20:27:20 INFO mapred.JobClient:   Map-Reduce Framework 

14/03/26 20:27:20 INFO mapred.JobClient:     Map output materialized bytes=164 

14/03/26 20:27:20 INFO mapred.JobClient:     Map input records=6 

14/03/26 20:27:20 INFO mapred.JobClient:     Reduce shuffle bytes=164 

14/03/26 20:27:20 INFO mapred.JobClient:     Spilled Records=24 

14/03/26 20:27:20 INFO mapred.JobClient:     Map output bytes=185 

14/03/26 20:27:20 INFO mapred.JobClient:     Total committed heap usage (bytes)=336338944 

14/03/26 20:27:20 INFO mapred.JobClient:     CPU time spent (ms)=2850 

14/03/26 20:27:20 INFO mapred.JobClient:     Combine input records=20 

14/03/26 20:27:20 INFO mapred.JobClient:     SPLIT_RAW_BYTES=218 

14/03/26 20:27:20 INFO mapred.JobClient:     Reduce input records=12 

14/03/26 20:27:20 INFO mapred.JobClient:     Reduce input groups=8 

14/03/26 20:27:20 INFO mapred.JobClient:     Combine output records=12 

14/03/26 20:27:20 INFO mapred.JobClient:     Physical memory (bytes) snapshot=464982016 

14/03/26 20:27:20 INFO mapred.JobClient:     Reduce output records=8 

14/03/26 20:27:20 INFO mapred.JobClient:     Virtual memory (bytes) snapshot=2182836224 

14/03/26 20:27:20 INFO mapred.JobClient:     Map output records=20 

</pre> 

最后，我们就可以去数据库里，查看统计的信息了，截图如下： 

 

至此，我们就完成了使用MR来读写数据库了，注意测试前，先确保自己的hadoop集群，可以正常工作。