一、window滑动窗口

1、概述



Spark Streaming提供了滑动窗口操作的支持,从而让我们可以对一个滑动窗口内的数据执行计算操作。每次掉落在窗口内的RDD的数据,
会被聚合起来执行计算操作,然后生成的RDD,会作为window DStream的一个RDD。比如下图中,就是对每三秒钟的数据执行一次滑动窗口计算,
这3秒内的3个RDD会被聚合起来进行处理,然后过了两秒钟,又会对最近三秒内的数据执行滑动窗口计算。所以每个滑动窗口操作,都必须指定
两个参数,窗口长度以及滑动间隔,而且这两个参数值都必须是batch间隔的整数倍。(Spark Streaming对滑动窗口的支持,是比Storm更加完善和强大的)



spark Steam优势_spark

 

2、window滑动窗口操作

 

spark Steam优势_spark Steam优势_02

 

案例:热点搜索词滑动统计,每隔10秒钟,统计最近60秒钟的搜索词的搜索频次,并打印出排名最靠前的3个搜索词以及出现次数

 

2、java案例



package cn.spark.study.streaming;

import java.util.List;

import org.apache.spark.SparkConf;
import org.apache.spark.api.java.JavaPairRDD;
import org.apache.spark.api.java.function.Function;
import org.apache.spark.api.java.function.Function2;
import org.apache.spark.api.java.function.PairFunction;
import org.apache.spark.streaming.Durations;
import org.apache.spark.streaming.api.java.JavaDStream;
import org.apache.spark.streaming.api.java.JavaPairDStream;
import org.apache.spark.streaming.api.java.JavaReceiverInputDStream;
import org.apache.spark.streaming.api.java.JavaStreamingContext;

import scala.Tuple2;

/**
 * 基于滑动窗口的热点搜索词实时统计
 * @author Administrator
 *
 */
public class WindowHotWord {
    
    public static void main(String[] args) {
        SparkConf conf = new SparkConf()
                .setMaster("local[2]")
                .setAppName("WindowHotWord");  
        JavaStreamingContext jssc = new JavaStreamingContext(conf, Durations.seconds(1));
        
        // 说明一下,这里的搜索日志的格式
        // leo hello
        // tom world
        JavaReceiverInputDStream<String> searchLogsDStream = jssc.socketTextStream("spark1", 9999);
        
        // 将搜索日志给转换成,只有一个搜索词,即可
        JavaDStream<String> searchWordsDStream = searchLogsDStream.map(new Function<String, String>() {

            private static final long serialVersionUID = 1L;

            @Override
            public String call(String searchLog) throws Exception {
                return searchLog.split(" ")[1];
            }
            
        });
        
        // 将搜索词映射为(searchWord, 1)的tuple格式
        JavaPairDStream<String, Integer> searchWordPairDStream = searchWordsDStream.mapToPair(
                
                new PairFunction<String, String, Integer>() {

                    private static final long serialVersionUID = 1L;

                    @Override
                    public Tuple2<String, Integer> call(String searchWord)
                            throws Exception {
                        return new Tuple2<String, Integer>(searchWord, 1);
                    }
                    
                });
        
        // 针对(searchWord, 1)的tuple格式的DStream,执行reduceByKeyAndWindow,滑动窗口操作
        // 第二个参数,是窗口长度,这里是60秒
        // 第三个参数,是滑动间隔,这里是10秒
        // 也就是说,每隔10秒钟,将最近60秒的数据,作为一个窗口,进行内部的RDD的聚合,然后统一对一个RDD进行后续
        // 计算
        // 所以说,这里的意思,就是,之前的searchWordPairDStream为止,其实,都是不会立即进行计算的
        // 而是只是放在那里
        // 然后,等待我们的滑动间隔到了以后,10秒钟到了,会将之前60秒的RDD,因为一个batch间隔是,5秒,所以之前
        // 60秒,就有12个RDD,给聚合起来,然后,统一执行redcueByKey操作
        // 所以这里的reduceByKeyAndWindow,是针对每个窗口执行计算的,而不是针对某个DStream中的RDD
        JavaPairDStream<String, Integer> searchWordCountsDStream =             
                 //Function2<T1, T2, R>:一个双参数函数,它接受类型为T1和T2的参数并返回一个R
                searchWordPairDStream.reduceByKeyAndWindow(new Function2<Integer, Integer, Integer>() {

                    private static final long serialVersionUID = 1L;                         

                    @Override
                    public Integer call(Integer v1, Integer v2) throws Exception {
                        return v1 + v2;
                    }
                    
                }, Durations.seconds(60), Durations.seconds(10));
        
        // 到这里为止,就已经可以做到,每隔10秒钟,出来,之前60秒的收集到的单词的统计次数
        // 执行transform操作,因为,一个窗口,就是一个60秒钟的数据,会变成一个RDD,然后,对这一个RDD
        // 根据每个搜索词出现的频率进行排序,然后获取排名前3的热点搜索词
        JavaPairDStream<String, Integer> finalDStream = searchWordCountsDStream.transformToPair(
                
                new Function<JavaPairRDD<String,Integer>, JavaPairRDD<String,Integer>>() {

                    private static final long serialVersionUID = 1L;

                    @Override
                    public JavaPairRDD<String, Integer> call(
                            JavaPairRDD<String, Integer> searchWordCountsRDD) throws Exception {
                        // 执行搜索词和出现频率的反转
                        JavaPairRDD<Integer, String> countSearchWordsRDD = searchWordCountsRDD
                                .mapToPair(new PairFunction<Tuple2<String,Integer>, Integer, String>() {

                                    private static final long serialVersionUID = 1L;

                                    @Override
                                    public Tuple2<Integer, String> call(
                                            Tuple2<String, Integer> tuple)
                                            throws Exception {
                                        return new Tuple2<Integer, String>(tuple._2, tuple._1);
                                    }
                                });
                        
                        // 然后执行降序排序
                        JavaPairRDD<Integer, String> sortedCountSearchWordsRDD = countSearchWordsRDD
                                .sortByKey(false);
                        
                        // 然后再次执行反转,变成(searchWord, count)的这种格式
                        JavaPairRDD<String, Integer> sortedSearchWordCountsRDD = sortedCountSearchWordsRDD
                                .mapToPair(new PairFunction<Tuple2<Integer,String>, String, Integer>() {

                                    private static final long serialVersionUID = 1L;

                                    @Override
                                    public Tuple2<String, Integer> call(
                                            Tuple2<Integer, String> tuple)
                                            throws Exception {
                                        return new Tuple2<String, Integer>(tuple._2, tuple._1);
                                    }
                                    
                                });
                        
                        // 然后用take(),获取排名前3的热点搜索词
                        List<Tuple2<String, Integer>> hogSearchWordCounts = 
                                sortedSearchWordCountsRDD.take(3);
                        for(Tuple2<String, Integer> wordCount : hogSearchWordCounts) {
                            System.out.println(wordCount._1 + ": " + wordCount._2);  
                        }
                        
                        return searchWordCountsRDD;
                    }
                      
                });
        
        // 这个无关紧要,只是为了触发job的执行,所以必须有output操作
        finalDStream.print();
        
        jssc.start();
        jssc.awaitTermination();
        jssc.close();
    }

}





##在eclipse中启动程序

##服务器上启动nc,并输入内容
[root@spark1 ~]# nc -lk 9999   
leo hello
tom word
leo hello
jack you
leo you

##统计结果
(hello,2)
(word,1)
(you,2)



 

3、scala案例



package cn.spark.study.streaming

import org.apache.spark.SparkConf
import org.apache.spark.streaming.StreamingContext
import org.apache.spark.streaming.Seconds

/**
 * @author Administrator
 */
object WindowHotWord {
  
  def main(args: Array[String]): Unit = {
    val conf = new SparkConf()
        .setMaster("local[2]")  
        .setAppName("WindowHotWord")
    val ssc = new StreamingContext(conf, Seconds(1))
    
    val searchLogsDStream = ssc.socketTextStream("spark1", 9999)  
    val searchWordsDStream = searchLogsDStream.map { _.split(" ")(1) }  
    val searchWordPairsDStream = searchWordsDStream.map { searchWord => (searchWord, 1) }  
    val searchWordCountsDSteram = searchWordPairsDStream.reduceByKeyAndWindow(
        (v1: Int, v2: Int) => v1 + v2, 
        Seconds(60), 
        Seconds(10))  
        
    val finalDStream = searchWordCountsDSteram.transform(searchWordCountsRDD => {
      val countSearchWordsRDD = searchWordCountsRDD.map(tuple => (tuple._2, tuple._1))  
      val sortedCountSearchWordsRDD = countSearchWordsRDD.sortByKey(false)  
      val sortedSearchWordCountsRDD = sortedCountSearchWordsRDD.map(tuple => (tuple._1, tuple._2))
      
      val top3SearchWordCounts = sortedSearchWordCountsRDD.take(3)
      for(tuple <- top3SearchWordCounts) {
        println(tuple)
      }
      
      searchWordCountsRDD
    })
    
    finalDStream.print()
    
    ssc.start()
    ssc.awaitTermination()
  }
  
}





##在eclipse中启动程序

##服务器上启动nc,并输入内容
[root@spark1 ~]# nc -lk 9999   
leo hello
leo hello
leo hello
leo word
leo word
leo word
leo hello
leo you
leo you

##统计结果
(hello,4)
(word,3)
(you,2)