Join应用
1、Reduce join
数据:
order.txt
id | pid | amount |
---|---|---|
1001 | 01 | 1 |
1002 | 02 | 2 |
1003 | 03 | 3 |
1004 | 01 | 4 |
1005 | 02 | 5 |
1006 | 03 | 6 |
pd.txt
pid | pname |
---|---|
01 | 小米 |
02 | 华为 |
03 | 格力 |
分析:
Map端的主要工作:为来自不同表或文件的key/value对,打标签以区别不同来源的记录。然后用连接字段作为key,其余部分和新加的标志作为value,最后进行输出。
Reduce端的主要工作:在Reduce端以连接字段作为key的分组已经完成,我们只需要在每一个分组当中将那些来源于不同文件的记录(在Map阶段已经打标志)分开,最后进行合并就ok了。
-
准备一个TableBean封装对象,使得能够适应以上两个表且区分两个表。
package com.hpu.review.rj; import org.apache.hadoop.io.Writable; import java.io.DataInput; import java.io.DataOutput; import java.io.IOException; public class TableBean implements Writable { private String id; private String pid; private Integer amount; private String pname; //区分两个表 private String flag; public TableBean(){} public String getId() { return id; } public void setId(String id) { this.id = id; } public String getPid() { return pid; } public void setPid(String pid) { this.pid = pid; } public Integer getAmount() { return amount; } public void setAmount(Integer amount) { this.amount = amount; } public String getPname() { return pname; } public void setPname(String pname) { this.pname = pname; } public String getFlag() { return flag; } public void setFlag(String flag) { this.flag = flag; } @Override public String toString() { //reduce阶段输出要用到,返回需要的字段 return id+"\t"+pname+"\t"+amount; } @Override public void write(DataOutput out) throws IOException { out.writeUTF(id); out.writeUTF(pid); out.writeInt(amount); out.writeUTF(pname); out.writeUTF(flag); } @Override public void readFields(DataInput in) throws IOException { this.id = in.readUTF(); this.pid = in.readUTF(); this.amount = in.readInt(); this.pname = in.readUTF(); this.flag = in.readUTF(); } }
-
写Mapper,首先继承Mapper且指定输出key为Text(pid),Value为TableBean对象;
-
通过context获取切片信息,由于InputSplite抽象类中没有文件路径及其文件名信息,将其强转为FileSplit,再取相应信息。
-
根据不同文件分别输出KV。
package com.hpu.review.rj; import org.apache.hadoop.io.LongWritable; import org.apache.hadoop.io.Text; import org.apache.hadoop.mapreduce.InputSplit; import org.apache.hadoop.mapreduce.Mapper; import org.apache.hadoop.mapreduce.lib.input.FileSplit; import java.io.IOException; public class MyMapper extends Mapper<LongWritable, Text, Text, TableBean> { private Text outK; private TableBean outV; private String name; @Override protected void setup(Context context) throws IOException, InterruptedException { outK = new Text(); outV = new TableBean(); //获取对应文件名称 InputSplit inputSplit = context.getInputSplit(); FileSplit fileSplit = (FileSplit)inputSplit; name = fileSplit.getPath().getName(); } @Override protected void map(LongWritable key, Text value, Context context) throws IOException, InterruptedException { String line = value.toString(); //判断是哪个文件,然后针对文件进行不同的操作 if (name.contains("order")){//订单表的处理 String[] orderInfo = line.split("\t"); outV.setId(orderInfo[0]); outV.setPid(orderInfo[1]); outV.setAmount(Integer.parseInt(orderInfo[2])); outV.setPname(""); outV.setFlag("order"); outK.set(orderInfo[1]); context.write(outK,outV); } else { //商品表的处理 String[] pdInfo = line.split("\t"); outV.setId(""); outV.setPid(pdInfo[0]); outV.setAmount(0); outV.setPname(pdInfo[1]); outV.setFlag("pd"); outK.set(pdInfo[0]); context.write(outK,outV); } } }
-
针对MyReducer,首先继承Reducer类;
-
指定Key为TableBean,Value为NullWritable;这里只把符合要求的bean对象中的pname替换即可;
-
values迭代器中,由于pid在pd表中是唯一的,因此在迭代器中也仅有一个,其余全部都是order。这也是为什么pdBean是一个对象,而tableBeans是一个集合的原因。
package com.hpu.review.rj; import org.apache.commons.beanutils.BeanUtils; import org.apache.hadoop.io.NullWritable; import org.apache.hadoop.io.Text; import org.apache.hadoop.mapreduce.Reducer; import java.io.IOException; import java.lang.reflect.InvocationTargetException; import java.util.ArrayList; public class MyReducer extends Reducer<Text,TableBean,TableBean, NullWritable> { //多个订单对应一个商品名 private ArrayList<TableBean> tableBeans; private TableBean pdBean; @Override protected void setup(Context context) throws IOException, InterruptedException { tableBeans = new ArrayList<>(); pdBean = new TableBean(); } @Override protected void reduce(Text key, Iterable<TableBean> values, Context context) throws IOException, InterruptedException { //每次都需要清空链表 tableBeans.clear(); for (TableBean value : values) { TableBean tmpBean = new TableBean(); //判断数据来自哪个表 if ("order".equals(value.getFlag())){ try { BeanUtils.copyProperties(tmpBean,value); //将临时TableBean对象添加到集合orderBeans tableBeans.add(tmpBean); } catch (IllegalAccessException e) { e.printStackTrace(); } catch (InvocationTargetException e) { e.printStackTrace(); } }else { try { BeanUtils.copyProperties(tmpBean,value); pdBean = tmpBean; } catch (IllegalAccessException e) { e.printStackTrace(); } catch (InvocationTargetException e) { e.printStackTrace(); } } } //遍历集合orderBeans,替换掉每个orderBean的pid为pname,然后写出 for (TableBean tableBean : tableBeans) { tableBean.setPname(pdBean.getPname()); context.write(tableBean,NullWritable.get()); } } }
-
Driver没什么好说的
package com.hpu.review.rj; import org.apache.hadoop.conf.Configuration; import org.apache.hadoop.fs.Path; import org.apache.hadoop.io.NullWritable; import org.apache.hadoop.io.Text; import org.apache.hadoop.mapreduce.Job; import org.apache.hadoop.mapreduce.lib.input.FileInputFormat; import org.apache.hadoop.mapreduce.lib.output.FileOutputFormat; import java.io.IOException; public class MyDriver { public static void main(String[] args) throws IOException, ClassNotFoundException, InterruptedException { Configuration conf = new Configuration(); Job job = Job.getInstance(conf); job.setJarByClass(MyDriver.class); job.setMapperClass(MyMapper.class); job.setReducerClass(MyReducer.class); job.setMapOutputKeyClass(Text.class); job.setMapOutputValueClass(TableBean.class); job.setOutputKeyClass(TableBean.class); job.setOutputValueClass(NullWritable.class); FileInputFormat.setInputPaths(job,new Path("E:\\Test\\input\\inputtable")); FileOutputFormat.setOutputPath(job,new Path("E:\\Test\\rj1")); job.waitForCompletion(true); } }
2、map join
采用DistributedCache
(1)在Mapper的setup阶段,将文件读取到缓存集合中。
(2)在Driver驱动类中加载缓存。
//缓存普通文件到Task运行节点。
job.addCacheFile(new URI("file:///e:/cache/pd.txt"));
//如果是集群运行,需要设置HDFS路径
job.addCacheFile(new URI("hdfs://hadoop102:8020/cache/pd.txt"));
-
先在MapJoinDriver驱动类中添加缓存文件
package com.hpu.review.mj; import org.apache.hadoop.conf.Configuration; import org.apache.hadoop.fs.Path; import org.apache.hadoop.io.NullWritable; import org.apache.hadoop.io.Text; import org.apache.hadoop.mapreduce.Job; import org.apache.hadoop.mapreduce.lib.input.FileInputFormat; import org.apache.hadoop.mapreduce.lib.output.FileOutputFormat; import java.io.IOException; import java.net.URI; import java.net.URISyntaxException; public class MyDriver { public static void main(String[] args) throws IOException, URISyntaxException, ClassNotFoundException, InterruptedException { Configuration conf = new Configuration(); Job job = Job.getInstance(conf); job.setJarByClass(MyDriver.class); job.setMapperClass(MyMapper.class); job.setNumReduceTasks(0); job.setMapOutputKeyClass(Text.class); job.setMapOutputValueClass(NullWritable.class); job.setOutputKeyClass(Text.class); job.setOutputValueClass(NullWritable.class); //添加pd.txt job.addCacheFile(new URI("file:///E:/Test/input/inputtablecache/pd.txt")); FileInputFormat.setInputPaths(job,new Path("E:\\Test\\input\\inputtable2")); FileOutputFormat.setOutputPath(job,new Path("E:\\Test\\mj1")); job.waitForCompletion(true); } }
-
在MapJoinMapper类中的setup方法中读取缓存文件
package com.hpu.review.mj; 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.LongWritable; import org.apache.hadoop.io.NullWritable; import org.apache.hadoop.io.Text; import org.apache.hadoop.mapreduce.Mapper; import java.io.BufferedReader; import java.io.IOException; import java.io.InputStreamReader; import java.net.URI; import java.util.HashMap; public class MyMapper extends Mapper<LongWritable, Text,Text, NullWritable> { private HashMap<String,String> pdmap = new HashMap<String,String>(); private Text text = new Text(); //任务开始前将pd数据缓存进pdMap @Override protected void setup(Context context) throws IOException, InterruptedException { //通过缓存文件得到小表数据pd.txt URI[] cacheFiles = context.getCacheFiles(); URI cacheFile = cacheFiles[0]; //开流 FileSystem fileSystem = FileSystem.get(context.getConfiguration()); FSDataInputStream open = fileSystem.open(new Path(cacheFile)); BufferedReader bufferedReader = new BufferedReader(new InputStreamReader(open, "UTF-8")); String line; while ((line = bufferedReader.readLine())!=null){ String[] split = line.split("\t"); pdmap.put(split[0],split[1]); } IOUtils.closeStreams(bufferedReader,open); } @Override protected void map(LongWritable key, Text value, Context context) throws IOException, InterruptedException { String[] line = value.toString().split("\t"); text.set(line[0]+"\t"+pdmap.get(line[1])+"\t"+line[2]); context.write(text,NullWritable.get()); } }
3、总结
Reduce Join方式中,合并的操作是在Reduce阶段完成,Reduce端的处理压力太大,Map节点的运算负载则很低,资源利用率不高,且在Reduce阶段极易产生数据倾斜。
解决方案:Map端实现数据合并。
Map Join适用于一张表十分小、一张表很大的场景。
在Map端缓存多张表,提前处理业务逻辑,这样增加Map端业务,减少Reduce端数据的压力,尽可能的减少数据倾斜。