前言

本章主要内容是讲述hadoop的分布式缓存的使用,通过分布式缓存可以将一些需要共享的数据在各个集群中共享。

准备工作

数据集:ufo-60000条记录,这个数据集有一系列包含下列字段的UFO目击事件记录组成,每条记录的字段都是以tab键分割,请看http://www.cnblogs.com/cafebabe-yun/p/8679994.html

  • sighting date:UFO目击事件发生时间
  • Recorded date:报告目击事件的时间
  • Location:目击事件发生的地点
  • Shape:UFO形状
  • Duration:目击事件持续时间
  • Dexcription:目击事件的大致描述

例子:

19950915 19950915 Redmond, WA 6 min. Young man w/ 2 co-workers witness tiny, distinctly white round disc drifting slowly toward NE. Flew in dir. 90 deg. to winds.

需要共享的数据:州名缩写与全称的对应关系

数据:

AL      Alabama
AK Alaska
AZ Arizona
AR Arkansas
CA California

Distributed Cache介绍

作用:使用分布式缓存,可以将map和reduce任务要用的通用只读文件在集群所有节点共享。

Distributed Cache的使用

题目:使用共享数据替换州名缩写

  • 将上面提到的共享数据保存为 states.txt 文件
  • 将states.txt文件上传到hadoop
hadoop dfs -put states.txt states.txt
  • 编写 UFORecordValidationMapper.java
import java.io.IOException;

import org.apache.hadoop.io.*;
import org.apache.hadoop.mapred.*;
import org.apache.hadoop.mapred.lib.*; public class UFORecordValidationMapper extends MapReduceBase implements Mapper<LongWritable, Text, LongWritable, Text> {
public void map(LongWritable key, Text value, OutputCollector<LongWritable, Text> output, Reporter reporter) throws IOException {
String line = value.toString();
if(validate(line)) {
output.collect(key, value);
}
} private boolean validate(String str) {
String[] parts = str.split("\t");
if(parts.length != 6) {
return false;
}
return true;
}
}
  • 编写 UFOLocation2.java
import java.io.*;
import java.util.*;
import java.net.*;
import java.util.regex.*; import org.apache.hadoop.conf.*;
import org.apache.hadoop.fs.Path;
import org.apache.hadoop.filecache.DistributedCache;
import org.apache.hadoop.io.*;
import org.apache.hadoop.mapred.*;
import org.apache.hadoop.mapred.lib.*; public class UFOLocation2 {
public static class MapClass extends MapReduceBase implements Mapper<LongWritable, Text, Text, LongWritable> {
private final static LongWritable one = new LongWritable(1);
private static Pattern locationPattern = Pattern.compile("[a-zA-Z]{2}[^a-zA-Z]*$");
private Map<String, String> stateNames; @Override
public void configure(JobConf job) {
try {
Path[] cacheFiles = DistributedCache.getLocalCacheFiles(job);
setupStateMap(cacheFiles[0].toString());
} catch (IOException e) {
System.err.println("Error reading state file.");
System.exit(1);
}
} private void setupStateMap(String fileName) throws IOException {
Map<String, String> stateCache = new HashMap<String, String>();
BufferedReader reader = new BufferedReader(new FileReader(fileName));
String line = null;
while((line = reader.readLine()) != null) {
String[] splits = line.split("\t");
stateCache.put(splits[0], splits[1]);
}
stateNames = stateCache;
} @Override
public void map(LongWritable key, Text value, OutputCollector<Text, LongWritable> output, Reporter reporter) throws IOException {
String line = value.toString();
String[] fields = line.split("\t");
String location = fields[2].trim();
if(location.length() >= 2) {
Matcher matcher = locationPattern.matcher(location);
if(matcher.find()) {
int start = matcher.start();
String state = location.substring(start, start + 2);
output.collect(new Text(lookupState(state.toUpperCase())), one);
}
}
} private String lookupState(String state) {
String fullName = stateNames.get(state);
if(fullName == null || "".equals(fullName)) {
fullName = state;
}
return fullName;
}
} public static void main(String...args) throws Exception {
Configuration config = new Configuration();
JobConf conf = new JobConf(config, UFOLocation2.class);
conf.setJobName("UFOLocation2");
DistributedCache.addCacheFile(new URI("/user/root/states.txt"), conf);
conf.setOutputKeyClass(Text.class);
conf.setOutputValueClass(LongWritable.class); JobConf mapconf1 = new JobConf(false);
ChainMapper.addMapper(conf, UFORecordValidationMapper.class, LongWritable.class, Text.class, LongWritable.class, Text.class, true, mapconf1);
JobConf mapconf2 = new JobConf(false);
ChainMapper.addMapper(conf, MapClass.class, LongWritable.class, Text.class, Text.class, LongWritable.class, true, mapconf2);
conf.setMapperClass(ChainMapper.class);
conf.setCombinerClass(LongSumReducer.class);
conf.setReducerClass(LongSumReducer.class); FileInputFormat.setInputPaths(conf, args[0]);
FileOutputFormat.setOutputPath(conf, new Path(args[1]));
JobClient.runJob(conf);
}
}
  • 编译上述两个文件
javac UFORecordValidationMapper.java UFOLocation2.java
  • 将编译好的文件打包成jar
jar cvf ufo.jar UFO*class
  • 提交打包好的jar包到hadoop上运行
hadoop jar ufo.jar UFOLocation2 ufo.tsv output
  • 从hadoop上获取结果到本地
hadoop dfs -get output/part-00000 ufo_result.txt
  • 查看结果
more ufo_result.txt

[hadoop](2) MapReducer:Distributed Cache的更多相关文章

  1. [转] .net core Session , Working with a distributed cache

    本文转自:https://docs.microsoft.com/en-us/aspnet/core/performance/caching/distributed By Steve Smith+ Di ...

  2. Distributed Cache Coherence at Scalable Requestor Filter Pipes that Accumulate Invalidation Acknowledgements from other Requestor Filter Pipes Using Ordering Messages from Central Snoop Tag

    A multi-processor, multi-cache system has filter pipes that store entries for request messages sent ...

  3. Hadoop之 MapReducer工作过程

    1. 从输入到输出 一个MapReducer作业经过了input,map,combine,reduce,output五个阶段,其中combine阶段并不一定发生,map输出的中间结果被分到reduce ...

  4. spark hadoop 对比 Resilient Distributed Datasets

    hadoop 迭代消耗大 每次迭代启动一个完整的MapReduce作业 spark 首要目标就是避免运算时 过多的网络和磁盘IO开销 Resilient Distributed Datasets ht ...

  5. Flink分布式缓存Distributed Cache

    1 分布式缓存 Flink提供了一个分布式缓存,类似于hadoop,可以使用户在并行函数中很方便的读取本地文件,并把它放在taskmanager节点中,防止task重复拉取. 此缓存的工作机制如下:程 ...

  6. Distributed Cache(分布式缓存)-SqlServer

    分布式缓存是由多个应用服务器共享的缓存,通常作为外部服务存储在单个应用服务器上,常用的有SqlServer,Redis,NCache. 分布式缓存可以提高ASP.NET Core应用程序的性能和可伸缩 ...

  7. hadoop系列四:mapreduce的使用(二)

    转载请在页首明显处注明作者与出处 一:说明 此为大数据系列的一些博文,有空的话会陆续更新,包含大数据的一些内容,如hadoop,spark,storm,机器学习等. 当前使用的hadoop版本为2.6 ...

  8. Hadoop官方文档翻译——MapReduce Tutorial

    MapReduce Tutorial(个人指导) Purpose(目的) Prerequisites(必备条件) Overview(综述) Inputs and Outputs(输入输出) MapRe ...

  9. hadoop常见问题汇集

    1 hadoop conf.addResource http://stackoverflow.com/questions/16017538/how-does-configuration-addreso ...

随机推荐

  1. Python解决NameError: name 'reload' is not defined的问题

    遇到这个问题,对于 Python 2.X: import sys reload(sys) sys.setdefaultencoding("utf-8") 对于 Python 3.3 ...

  2. Learn Python the hard way, ex45 对象、类、以及从属关系

    #!/usr/bin/python #coding:utf-8 # animal is-a object(yes,sort of sonfusing)look at the extra credit ...

  3. 69.x的平方根

    class Solution: def mySqrt(self, x: int) -> int: if x < 2: return x left, right = 1, x//2 whil ...

  4. idea 社区版本创建javaweb项目 使用tomcat

    1.创建maven  webapp项目 2.pom文件添加依赖及tomcat7-maven-plugin插件 <dependencies> <dependency> <g ...

  5. Java对象多态性——对象的转型(引用类型之间的类型转换)

    Java引用变量有两个类型:编译时类型和运行时类型.编译时类型由声明该变量时使用的类型决定,运行时类型由实际赋给该变量的对象决定. 若编译时类型和运行时类型不一致,就出现多态 正常的方法调用(本态调用 ...

  6. POJ-1064.Cablemaster.(二分法枚举值求最优值)

    题目链接 本题大意:给你n个长度为value[ i ]的长木板,让你切割成为等长的k份,问你切割的最大长度是多少. 本题思路:其实很容易可以想到先找到一个上界和一个下界,开始枚举里面的所有长度,取最长 ...

  7. Java JNA (四)—— void**、void*、char**、char*、int*等类型映射关系及简单示例

    ByReference类有很多子类,这些类都非常有用. ByteByReference.DoubleByReference.FloatByReference. IntByReference.LongB ...

  8. ubuntu下载地址

    http://mirrors.aliyun.com/ubuntu-releases/16.04/

  9. ASE Code Search

    重现基线模型 Hamel's model 基线模型原理 如何实现semantic search?在已有数据库的基础上,衡量一个句子和每段代码的相关性再进行排序,选出最优代码片段即可实现一个通用的cod ...

  10. 树——minimum-depth-of-binary-tree(二叉树的最小深度)

    问题: Given a binary tree, find its minimum depth.The minimum depth is the number of nodes along the s ...