一、MapJoin-DistributedCache 应用

1、mapreduce join 介绍

在各种实际业务场景中,按照某个关键字对两份数据进行连接是非常常见的。如果两份数据 都比较小,那么可以直接在内存中完成连接。如果是大数据量的呢? 显然,在内存中进行连 接会发生 OOM。 MapReduce 可以用来解决大数据量的链接
MapReduce 的 Join 操作主要分两类: MapJoin 和 ReduceJoin

先看 ReduceJoin:
(1)map 阶段,两份数据 data1 和 data2 会被 map 分别读入,解析成以链接字段为 key 以查 询字段为 value 的 key-value 对,并标明数据来源是 data1 还是 data2。
(2)reduce 阶段, reducetask 会接收来自 data1 和 data2 的相同 key 的数据,在 reduce 端进 行乘积链接, 最直接的影响是很消耗内存,导致 OOM

再看 MapJoin:
MapJoin 适用于有一份数据较小的连接情况。做法是直接把该小份数据直接全部加载到内存 当中,按链接关键字建立索引。 然后大份数据就作为 MapTask 的输入,对 map()方法的每次 输入都去内存当中直接去匹配连接。 然后把连接结果按 key 输出,这种方法要使用 hadoop
中的 DistributedCache 把小份数据分布到各个计算节点,每个 maptask 执行任务的节点都需 要加载该数据到内存,并且按连接关键字建立索引 
(map读的是大表数据,在读大表之前,把小表数据放到内存当中,用setup方法)

2、需求

现有两份数据 movies.dat 和 ratings.dat 数据样式分别为:

Movies.dat:
     1::Toy Story (1995)::Animation|Children's|Comedy
     2::Jumanji (1995)::Adventure|Children's|Fantasy
     3::Grumpier Old Men (1995)::Comedy|Romance
     字段含义: movieid, moviename, movietype

Ratings.dat
    1::1193::5::978300760
    1::661::3::978302109
    1::914::3::978301968

字段含义: userid, movieid, rate, timestamp

现要求对两表进行连接,要求输出最终的结果有以上六个字段:
movieid, userid, rate, moviename, movietype, timestamp

3、实现

第一步:封装 MovieRate,方便数据的排序和序列化

package com.ghgj.mr.mymapjoin;
import java.io.DataInput;
import java.io.DataOutput;
import java.io.IOException;
import org.apache.hadoop.io.WritableComparable;
public class MovieRate implements WritableComparable<MovieRate>{
private String movieid;
private String userid;
private int rate;
private String movieName;
private String movieType;
private long ts;
public String getMovieid() {
return movieid;
}
public void setMovieid(String movieid) {
this.movieid = movieid;
}
public String getUserid() {
return userid;
}
public void setUserid(String userid) {
this.userid = userid;
}
public int getRate() {
return rate;
}
public void setRate(int rate) {
this.rate = rate;
}
public String getMovieName() {
return movieName;
}
public void setMovieName(String movieName) {
this.movieName = movieName;
}
public String getMovieType() {
return movieType;
}
public void setMovieType(String movieType) {
this.movieType = movieType;
}
public long getTs() {
return ts;
}
public void setTs(long ts) {
this.ts = ts;
}
public MovieRate() {
}
public MovieRate(String movieid, String userid, int rate, String movieName,
String movieType, long ts) {
this.movieid = movieid;
this.userid = userid;
this.rate = rate;
this.movieName = movieName;
this.movieType = movieType;
this.ts = ts;
}
@Override
public String toString() {
return movieid + "\t" + userid + "\t" + rate + "\t" + movieName
+ "\t" + movieType + "\t" + ts;
}
@Override
public void write(DataOutput out) throws IOException {
out.writeUTF(movieid);
out.writeUTF(userid);
out.writeInt(rate);
out.writeUTF(movieName);
out.writeUTF(movieType);
out.writeLong(ts);
}
@Override
public void readFields(DataInput in) throws IOException {
this.movieid = in.readUTF();
this.userid = in.readUTF();
this.rate = in.readInt();
this.movieName = in.readUTF();
this.movieType = in.readUTF();
this.ts = in.readLong();
}
@Override
public int compareTo(MovieRate mr) {
int it = mr.getMovieid().compareTo(this.movieid);
if(it == 0){
return mr.getUserid().compareTo(this.userid);
}else{
return it;
}
}
}

第二步:编写mapreduce程序

package com.ghgj.mr.mymapjoin;
import java.io.BufferedReader;
import java.io.FileReader;
import java.io.IOException;
import java.net.URI;
import java.util.HashMap;
import java.util.Map;
import org.apache.hadoop.conf.Configuration;
import org.apache.hadoop.fs.FileSystem;
import org.apache.hadoop.fs.Path;
import org.apache.hadoop.io.IOUtils;
import org.apache.hadoop.io.LongWritable;
import org.apache.hadoop.io.NullWritable;
import org.apache.hadoop.io.Text;
import org.apache.hadoop.mapreduce.Job;
import org.apache.hadoop.mapreduce.Mapper;
import org.apache.hadoop.mapreduce.Reducer;
import org.apache.hadoop.mapreduce.filecache.DistributedCache;
import org.apache.hadoop.mapreduce.lib.input.FileInputFormat;
import org.apache.hadoop.mapreduce.lib.output.FileOutputFormat;
public class MovieRatingMapJoinMR {
public static void main(String[] args) throws Exception {
Configuration conf = new Configuration();
conf.set("fs.defaultFS", "hdfs://hadoop02:9000");
System.setProperty("HADOOP_USER_NAME","hadoop");

Job job = Job.getInstance(conf);
// job.setJarByClass(MovieRatingMapJoinMR.class);
job.setJar("/home/hadoop/mrmr.jar");
job.setMapperClass(MovieRatingMapJoinMRMapper.class);
job.setMapOutputKeyClass(MovieRate.class);
job.setMapOutputValueClass(NullWritable.class);
// job.setReducerClass(MovieRatingMapJoinMReducer.class);
// job.setOutputKeyClass(MovieRate.class);
// job.setOutputValueClass(NullWritable.class);
job.setNumReduceTasks(0);
String minInput = args[0];
String maxInput = args[1];
String output = args[2];
FileInputFormat.setInputPaths(job, new Path(maxInput));
Path outputPath = new Path(output);
FileSystem fs = FileSystem.get(conf);
if(fs.exists(outputPath)){
fs.delete(outputPath, true);
}
FileOutputFormat.setOutputPath(job, outputPath);
URI uri = new Path(minInput).toUri();
job.addCacheFile(uri);
boolean status = job.waitForCompletion(true);
System.exit(status?0:1);
}

  

二、自定义 OutputFormat—数据分类输出

实现:自定义 OutputFormat,改写其中的 RecordWriter,改写具体输出数据的方法 write()

package com.ghgj.mr.score_outputformat;

import java.io.IOException;

import org.apache.hadoop.conf.Configuration;
import org.apache.hadoop.fs.FSDataOutputStream;
import org.apache.hadoop.fs.FileSystem;
import org.apache.hadoop.fs.Path;
import org.apache.hadoop.io.IOUtils;
import org.apache.hadoop.io.NullWritable;
import org.apache.hadoop.io.Text;
import org.apache.hadoop.mapreduce.RecordWriter;
import org.apache.hadoop.mapreduce.TaskAttemptContext;
import org.apache.hadoop.mapreduce.lib.output.TextOutputFormat; public class MyScoreOutputFormat extends TextOutputFormat<Text, NullWritable>{ @Override
public RecordWriter<Text, NullWritable> getRecordWriter(
TaskAttemptContext job) throws IOException, InterruptedException {
Configuration configuration = job.getConfiguration(); FileSystem fs = FileSystem.get(configuration);
Path p1 = new Path("/score1/outpu1");
Path p2 = new Path("/score2/outpu2"); if(fs.exists(p1)){
fs.delete(p1, true);
}
if(fs.exists(p2)){
fs.delete(p2, true);
} FSDataOutputStream fsdout1 = fs.create(p1);
FSDataOutputStream fsdout2 = fs.create(p2);
return new MyRecordWriter(fsdout1, fsdout2);
} static class MyRecordWriter extends RecordWriter<Text, NullWritable>{ FSDataOutputStream dout1 = null;
FSDataOutputStream dout2 = null; public MyRecordWriter(FSDataOutputStream dout1, FSDataOutputStream dout2) {
super();
this.dout1 = dout1;
this.dout2 = dout2;
} @Override
public void write(Text key, NullWritable value) throws IOException,
InterruptedException {
// TODO Auto-generated method stub String[] strs = key.toString().split("::");
if(strs[0].equals("1")){
dout1.writeBytes(strs[1]+"\n");
}else{
dout2.writeBytes(strs[1]+"\n");
}
} @Override
public void close(TaskAttemptContext context) throws IOException,
InterruptedException {
IOUtils.closeStream(dout2);
IOUtils.closeStream(dout1);
}
}
}

  

package com.ghgj.mr.score_outputformat;

import java.io.IOException;

import org.apache.hadoop.conf.Configuration;
import org.apache.hadoop.conf.Configured;
import org.apache.hadoop.fs.FileSystem;
import org.apache.hadoop.fs.Path;
import org.apache.hadoop.io.LongWritable;
import org.apache.hadoop.io.NullWritable;
import org.apache.hadoop.io.Text;
import org.apache.hadoop.mapreduce.Job;
import org.apache.hadoop.mapreduce.Mapper;
import org.apache.hadoop.mapreduce.lib.input.FileInputFormat;
import org.apache.hadoop.mapreduce.lib.input.TextInputFormat;
import org.apache.hadoop.mapreduce.lib.output.FileOutputFormat;
import org.apache.hadoop.util.Tool;
import org.apache.hadoop.util.ToolRunner; public class ScoreOutputFormatMR extends Configured implements Tool{ // 这个run方法就相当于Driver
@Override
public int run(String[] args) throws Exception { Configuration conf = new Configuration();
conf.set("fs.defaultFS", "hdfs://hadoop02:9000");
System.setProperty("HADOOP_USER_NAME", "hadoop");
Job job = Job.getInstance(conf); job.setMapperClass(ScoreOutputFormatMRMapper.class);
job.setMapOutputKeyClass(Text.class);
job.setMapOutputValueClass(NullWritable.class); job.setNumReduceTasks(0); // 这就是默认的输入输出组件
job.setInputFormatClass(TextInputFormat.class);
// 这是默认往外输出数据的组件
// job.setOutputFormatClass(TextOutputFormat.class);
job.setOutputFormatClass(MyScoreOutputFormat.class); FileInputFormat.setInputPaths(job, new Path("/scorefmt"));
Path output = new Path("/scorefmt/output");
FileSystem fs = FileSystem.get(conf);
if(fs.exists(output)){
fs.delete(output, true);
}
FileOutputFormat.setOutputPath(job, output); boolean status = job.waitForCompletion(true);
return status?0:1;
} public static void main(String[] args) throws Exception { int run = new ToolRunner().run(new ScoreOutputFormatMR(), args);
System.exit(run);
} static class ScoreOutputFormatMRMapper extends Mapper<LongWritable, Text, Text, NullWritable>{
@Override
protected void map(LongWritable key, Text value,
Mapper<LongWritable, Text, Text, NullWritable>.Context context)
throws IOException, InterruptedException { String[] split = value.toString().split("\t");
if(split.length-2 >= 6){
context.write(new Text("1::"+value.toString()), NullWritable.get());
}else{
context.write(new Text("2::"+value.toString()), NullWritable.get());
}
}
}
}

三、自定义 InputFormat—小文件合并

第一步:自定义InputFormat

package com.ghgj.mr.format.input;

import java.io.IOException;

import org.apache.hadoop.fs.Path;
import org.apache.hadoop.io.NullWritable;
import org.apache.hadoop.io.Text;
import org.apache.hadoop.mapreduce.InputSplit;
import org.apache.hadoop.mapreduce.JobContext;
import org.apache.hadoop.mapreduce.RecordReader;
import org.apache.hadoop.mapreduce.TaskAttemptContext;
import org.apache.hadoop.mapreduce.lib.input.FileInputFormat; public class WholeFileInputFormat extends FileInputFormat<NullWritable, Text> {
// 设置每个小文件不可分片,保证一个小文件生成一个key-value键值对
@Override
protected boolean isSplitable(JobContext context, Path file) {
return false;
} @Override
public RecordReader<NullWritable, Text> createRecordReader(
InputSplit split, TaskAttemptContext context) throws IOException,
InterruptedException {
WholeFileRecordReader reader = new WholeFileRecordReader();
reader.initialize(split, context);
return reader;
}
}

  第二步:编写自定义的 RecordReader

package com.ghgj.mr.format.input;

import java.io.IOException;

import org.apache.hadoop.conf.Configuration;
import org.apache.hadoop.fs.FSDataInputStream;
import org.apache.hadoop.fs.FileSystem;
import org.apache.hadoop.fs.Path;
import org.apache.hadoop.io.IOUtils;
import org.apache.hadoop.io.NullWritable;
import org.apache.hadoop.io.Text;
import org.apache.hadoop.mapreduce.InputSplit;
import org.apache.hadoop.mapreduce.RecordReader;
import org.apache.hadoop.mapreduce.TaskAttemptContext;
import org.apache.hadoop.mapreduce.lib.input.FileSplit; class WholeFileRecordReader extends RecordReader<NullWritable, Text> {
private FileSplit fileSplit;
private Configuration conf;
private Text value = new Text();
private boolean processed = false; @Override
public void initialize(InputSplit split, TaskAttemptContext context)
throws IOException, InterruptedException {
this.fileSplit = (FileSplit) split;
this.conf = context.getConfiguration();
} @Override
public boolean nextKeyValue() throws IOException, InterruptedException {
if (!processed) {
// 获取 输入逻辑切片的 字节数组
byte[] contents = new byte[(int) fileSplit.getLength()];
// 通过 filesplit获取该逻辑切片在文件系统的位置
Path file = fileSplit.getPath();
FileSystem fs = file.getFileSystem(conf);
FSDataInputStream in = null;
try {
// 文件系统对象fs打开一个file的输入流
in = fs.open(file);
/**
* in是输入流
* contents是存这个流读取的到数的数据的字节数组
*
*/
IOUtils.readFully(in, contents, 0, contents.length); value.set(contents, 0, contents.length); } finally {
IOUtils.closeStream(in);
}
processed = true;
return true;
}
return false;
} @Override
public NullWritable getCurrentKey() throws IOException, InterruptedException {
return NullWritable.get();
} @Override
public Text getCurrentValue() throws IOException, InterruptedException {
return value;
} @Override
public float getProgress() throws IOException {
return processed ? 1.0f : 0.0f;
} @Override
public void close() throws IOException {
// do nothing
}
}

   第三步:编写mapreduce程序

package com.ghgj.mr.format.input;

import java.io.IOException;

import org.apache.hadoop.conf.Configuration;
import org.apache.hadoop.conf.Configured;
import org.apache.hadoop.fs.FileSystem;
import org.apache.hadoop.fs.Path;
import org.apache.hadoop.io.NullWritable;
import org.apache.hadoop.io.Text;
import org.apache.hadoop.mapreduce.InputSplit;
import org.apache.hadoop.mapreduce.Job;
import org.apache.hadoop.mapreduce.Mapper;
import org.apache.hadoop.mapreduce.Reducer;
import org.apache.hadoop.mapreduce.lib.input.FileInputFormat;
import org.apache.hadoop.mapreduce.lib.input.FileSplit;
import org.apache.hadoop.mapreduce.lib.input.TextInputFormat;
import org.apache.hadoop.mapreduce.lib.output.FileOutputFormat;
import org.apache.hadoop.util.Tool;
import org.apache.hadoop.util.ToolRunner; public class SmallFilesConvertToBigMR extends Configured implements Tool { public static void main(String[] args) throws Exception {
int exitCode = ToolRunner.run(new SmallFilesConvertToBigMR(), args);
System.exit(exitCode);
} static class SmallFilesConvertToBigMRMapper extends
Mapper<NullWritable, Text, Text, Text> { private Text filenameKey;
@Override
protected void setup(Context context) throws IOException,
InterruptedException {
InputSplit split = context.getInputSplit();
Path path = ((FileSplit) split).getPath();
filenameKey = new Text(path.toString());
} @Override
protected void map(NullWritable key, Text value, Context context)
throws IOException, InterruptedException {
context.write(filenameKey, value);
}
} static class SmallFilesConvertToBigMRReducer extends
Reducer<Text, Text, NullWritable, Text> {
@Override
protected void reduce(Text filename, Iterable<Text> bytes,
Context context) throws IOException, InterruptedException {
context.write(NullWritable.get(), bytes.iterator().next());
}
} @Override
public int run(String[] args) throws Exception {
Configuration conf = new Configuration();
conf.set("fs.defaultFS", "hdfs://hadoop02:9000");
System.setProperty("HADOOP_USER_NAME", "hadoop");
Job job = Job.getInstance(conf, "combine small files to bigfile"); job.setJarByClass(SmallFilesConvertToBigMR.class); job.setMapOutputKeyClass(Text.class);
job.setMapOutputValueClass(Text.class);
job.setMapperClass(SmallFilesConvertToBigMRMapper.class); job.setOutputKeyClass(NullWritable.class);
job.setOutputValueClass(Text.class);
job.setReducerClass(SmallFilesConvertToBigMRReducer.class); // TextInputFormat是默认的数据读取组件
// job.setInputFormatClass(TextInputFormat.class);
// 不是用默认的读取数据的Format,我使用自定义的 WholeFileInputFormat
job.setInputFormatClass(WholeFileInputFormat.class); Path input = new Path("/smallfiles");
Path output = new Path("/bigfile");
FileInputFormat.setInputPaths(job, input);
FileSystem fs = FileSystem.get(conf);
if (fs.exists(output)) {
fs.delete(output, true);
}
FileOutputFormat.setOutputPath(job, output); int status = job.waitForCompletion(true) ? 0 : 1;
return status;
}
}

  

MapReduce(四) 典型编程场景(二)的更多相关文章

  1. 《Data-Intensive Text Processing with mapReduce》读书笔记之二:mapreduce编程、框架及运行

    搜狐视频的屌丝男士第二季大结局了,惊现波多野老师,怀揣着无比鸡冻的心情啊,可惜随着剧情的推进发展,并没有出现期待中的屌丝奇遇,大鹏还是没敢冲破尺度的界线.想百度些种子吧,又不想让电脑留下污点证据,要知 ...

  2. SSD固态盘应用于Ceph集群的四种典型使用场景

    在虚拟化及云计算技术大规模应用于企业数据中心的科技潮流中,存储性能无疑是企业核心应用是否虚拟化.云化的关键指标之一.传统的做法是升级存储设备,但这没解决根本问题,性能和容量不能兼顾,并且解决不好设备利 ...

  3. 从Paxos到ZooKeeper-三、ZooKeeper的典型应用场景

    ZooKeeper是一个典型的发布/订阅模式的分布式数据管理与协调框架,开发人员可以使用它来进行分布式数据的发布与订阅.另一方面,通过对ZooKeeper中丰富的数据节点类型进行交叉使用,配合Watc ...

  4. 并发编程(二):全视角解析volatile

    一.目录 1.引入话题-发散思考 2.volatile深度解析 3.解决volatile原子性问题 4.volatile应用场景 二.引入话题-发散思考 public class T1 { /*vol ...

  5. 并发编程(二)concurrent 工具类

    并发编程(二)concurrent 工具类 一.CountDownLatch 经常用于监听某些初始化操作,等初始化执行完毕后,通知主线程继续工作. import java.util.concurren ...

  6. 搞懂分布式技术6:Zookeeper典型应用场景及实践

    搞懂分布式技术6:Zookeeper典型应用场景及实践 一.ZooKeeper典型应用场景实践 ZooKeeper是一个高可用的分布式数据管理与系统协调框架.基于对Paxos算法的实现,使该框架保证了 ...

  7. 基于Apache Hudi构建数据湖的典型应用场景介绍

    1. 传统数据湖存在的问题与挑战 传统数据湖解决方案中,常用Hive来构建T+1级别的数据仓库,通过HDFS存储实现海量数据的存储与水平扩容,通过Hive实现元数据的管理以及数据操作的SQL化.虽然能 ...

  8. 《高性能javascript》 领悟随笔之-------DOM编程篇(二)

    <高性能javascript> 领悟随笔之-------DOM编程篇二 序:在javaSctipt中,ECMASCRIPT规定了它的语法,BOM实现了页面与浏览器的交互,而DOM则承载着整 ...

  9. 【转】《深入理解计算机系统》C程序中常见的内存操作有关的典型编程错误

    原文地址:http://blog.csdn.net/slvher/article/details/9150597 对C/C++程序员来说,内存管理是个不小的挑战,绝对值得慎之又慎,否则让由上万行代码构 ...

随机推荐

  1. 跟浩哥学自动化测试Selenium -- Selenium简介 (1)

    Selenium 简介 Selenium 是一款开源的web自动化测试工具,用来模拟对浏览器的操作(主要是对页面元素的操作),简单来讲,其实就是一个jar包.Selenium早期的版本比如1.0市场占 ...

  2. AWS探索及创建一个aws EC2实例

    一.AWS登陆 1.百度搜索aws,或者浏览器输入:http://aws.amazon.com 2.输入账户及密码登陆(注册流程按照提示走即可) 二.创建EC2实例(相当于阿里云的ecs) 1.找到E ...

  3. 前端基础HTML

    web的服务本质 浏览器发送请求>>>HTTP协议>>>服务端接受请求>>>服务端返回响应>>>服务端把HTML文件内容发给浏览 ...

  4. Javascript深入__proto__和prototype的区别和联系

    有一个一个装逼的同事,写了一段代码 function a(){} a.__proto__.__proto__.__proto__ 然后问我,下面这个玩意a.__proto__.__proto__.__ ...

  5. Spring单元测试集成H2数据库

    项目源代码在:Spring-H2测试 H2简介 H2数据库是一种由Java编写的,极小,速度极快,可嵌入式的数据库.非常适合用在单元测试等数据不需要保存的场景下面. 以下时其官网的介绍: {% blo ...

  6. RyuBook1.0案例三:REST Linkage

    REST Linkage 该小结主要介绍如何添加一个REST Link 函数 RYU本身提供了一个类似WSGI的web服务器功能.借助这个功能,我们可以创建一个REST API. 基于创建的REST ...

  7. cinder的组件

    跟nova相似,cinder也有很多组件,每个组件负责各自的业务,然后共同协作完成volume的管理.组件之间的通信方式与nova个组件之间的通信方式相同,都是通过消息队列进行通信. cinder-a ...

  8. POJ 1417 并查集 dp

    After having drifted about in a small boat for a couple of days, Akira Crusoe Maeda was finally cast ...

  9. 第17次Scrum会议(10/29)【欢迎来怼】

    一.小组信息 队名:欢迎来怼小组成员队长:田继平成员:李圆圆,葛美义,王伟东,姜珊,邵朔,冉华 小组照片 二.开会信息 时间:2017/10/29 17:20~17:42,总计22min.地点:东北师 ...

  10. U盘安装OSX

    1.插入U盘,磁盘工具,格式化U盘为Mac OS X拓展 (日志式): 2.去网站搜索recovery disk assistant,此文件大约1.1M,直接打开使用它制作启动盘,进度条完毕就完成了. ...