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。

分析结果 一般异步写入DB的比较多

同步的话，压力太大可能扛不住。

确实如此